api-20261104T020710Z / documentation capture

about-claude/models/optimizing-for-cost-and-intelligence Changed · +5 / -3 lines

from line 49

4949

5050#### Why caching comes first

5151

52Turn on [prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) before any other lever, because every turn of an agentic task resends the entire growing conversation: system prompt, tool definitions, and every prior turn.

52Turn on [prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) before any other lever, because by default every turn of an agentic task resends the entire growing conversation: system prompt, tool definitions, and every prior turn.

5353

5454* A 40-turn task sends its first turn 40 times, so task cost grows with roughly the square of turn count.

5555* Caching does not stop the resending, but each resend costs about a tenth as much (a twentieth on Claude Opus 5.5 and Claude Sonnet 5.5, a fortieth on Claude Fable 5.1 and Claude Mythos 5.1) and processes faster.

5656* The prefix is billed at the [cache-read rate](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#pricing) instead of the input price, and each turn pays the 1.25x cache-write rate only for what is new.

5757

58[Message threads](https://platform.claude.com/docs/en/build-with-claude/message-threads), in beta, stop the resending while a thread lasts, until 5 minutes after its last request. They lower time to first token, not cost: the model still reads the whole conversation, so keep `cache_control` on every request. Threads don’t support compaction or most context editing.

59

5860**What good looks like.**

5961

6062* Over a full day of real traffic, agent loops read a median 84% of their input from the cache, and the top 10% of them, coding or not, read 94% or more[17](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs).

from line 170

168170 jq '.max_tokens = 0 | del(.stream)' last_request.json | \

169171 curl https://api.anthropic.com/v1/messages \

170172 -H "x-api-key: $ANTHROPIC_API_KEY" \

171 -H "anthropic-version: 2023-06-01" \

173 -H "anthropic-version: 2026-10-21" \

172174 -H "content-type: application/json" \

173175 --data-binary @-

174176 ```

from line 314

312314Route every request no one is waiting on through the [Batch API](https://platform.claude.com/docs/en/build-with-claude/batch-processing), and keep the interactive path for the rest. The Batch API takes 50% off every token of a request, including cached ones, in exchange for results arriving any time within 24 hours.

313315

314316* Batching is the second-largest free lever after caching for unattended agent work: evaluation runs, backfills, and scheduled jobs such as a recurring run of the issue-triage agent from the [token-trimming measurement](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#trim-input-and-context-tokens).

315* Batching combines with most of this page, but not with [fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode) or Claude Managed Agents sessions (see [Claude Managed Agents pricing](https://platform.claude.com/docs/en/about-claude/pricing#claude-managed-agents-pricing)).

317* Batching combines with most of this page, but not with [fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode), message threads, or Claude Managed Agents sessions (see [Claude Managed Agents pricing](https://platform.claude.com/docs/en/about-claude/pricing#claude-managed-agents-pricing)).

316318

317319### Audit prompts against the current model

318320

about-claude/pricing Changed · +8 / -8 lines

from line 24

from line 57

5757

5858This section covers partner-operated cloud platforms, where the cloud provider invoices you. For Anthropic-operated cloud platforms billed through a marketplace, see [Claude Platform on AWS pricing](https://platform.claude.com/docs/en/about-claude/pricing#claude-platform-on-aws-pricing), [Claude Models on Google Cloud (Anthropic managed) billing](https://platform.claude.com/docs/en/build-with-claude/claude-models-on-google-cloud/anthropic-managed#billing), and [Claude in Microsoft Foundry pricing](https://platform.claude.com/docs/en/about-claude/pricing#claude-in-microsoft-foundry-pricing).

5959

60Claude models are available on [Amazon Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock) and [Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/build-with-claude/claude-models-on-google-cloud/google-managed). For official pricing, visit:

60For official pricing on [Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/build-with-claude/claude-models-on-google-cloud/google-managed) and [Amazon Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock), visit:

6161

6262* [Amazon Bedrock pricing](https://aws.amazon.com/bedrock/pricing/)

6363* [Google Cloud pricing](https://cloud.google.com/vertex-ai/generative-ai/pricing#claude-models)

from line 159

159159

160160For Claude 4.6 and later models, specifying US-only inference through the `inference_geo` parameter incurs a 1.1x multiplier on all token pricing categories, including input tokens, output tokens, cache writes, and cache reads. On Claude Haiku 5.5, the multiplier also applies to the higher prices for prompts over 100,000 tokens (see [Long context pricing](https://platform.claude.com/docs/en/about-claude/pricing#long-context-pricing)). Global routing (the default) uses standard pricing.

161161

162This applies to the Claude API (first-party), Claude Platform on AWS, and Claude Models on Google Cloud (Anthropic managed). On Claude in Microsoft Foundry, the same 1.1x multiplier applies to deployments that use the US Data Zone Standard deployment type (see [Inference geography](https://platform.claude.com/docs/en/about-claude/pricing#foundry-inference-geography)). Partner-operated platforms, Bedrock and Claude Models on Google Cloud (Google managed), have independent regional pricing. See [Bedrock pricing](https://aws.amazon.com/bedrock/pricing/) and [Google Cloud pricing](https://cloud.google.com/vertex-ai/generative-ai/pricing#claude-models) for details. Earlier models do not support the `inference_geo` parameter and always use standard pricing; requests that include the parameter on these models return a 400 error.

162This applies to the Claude API (first-party), Claude Platform on AWS, and Claude Models on Google Cloud (Anthropic managed). On Claude in Microsoft Foundry, the same 1.1x multiplier applies to deployments that use the US Data Zone Standard deployment type (see [Inference geography](https://platform.claude.com/docs/en/about-claude/pricing#foundry-inference-geography)). Partner-operated platforms, Claude Models on Google Cloud (Google managed) and Bedrock, have independent regional pricing. See [Bedrock pricing](https://aws.amazon.com/bedrock/pricing/) and [Google Cloud pricing](https://cloud.google.com/vertex-ai/generative-ai/pricing#claude-models) for details. Earlier models do not support the `inference_geo` parameter and always use standard pricing; requests that include the parameter on these models return a 400 error.

163163

164164For more information, see [Data residency](https://platform.claude.com/docs/en/manage-claude/data-residency).

165165

from line 197

197197| Claude Opus 4.7 | $2.50 / MTok | $12.50 / MTok |

198198| Claude Opus 4.6 | $2.50 / MTok | $12.50 / MTok |

199199| Claude Opus 4.5 | $2.50 / MTok | $12.50 / MTok |

200| Claude Opus 4.1 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $7.50 / MTok | $37.50 / MTok |

200| Claude Opus 4.1 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $7.50 / MTok | $37.50 / MTok |

201201| Claude Opus 4 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $7.50 / MTok | $37.50 / MTok |

202202| Claude Sonnet 5.5 | $1 / MTok | $5 / MTok |

203203| Claude Sonnet 5 | $1 / MTok | $5 / MTok |

204204| Claude Sonnet 4.6 | $1.50 / MTok | $7.50 / MTok |

205205| Claude Sonnet 4.5 | $1.50 / MTok | $7.50 / MTok |

206| Claude Sonnet 4 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $1.50 / MTok | $7.50 / MTok |

206| Claude Sonnet 4 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $1.50 / MTok | $7.50 / MTok |

207207| Claude Haiku 5.5 (for prompts up to 100,000 tokens) | $0.05 / MTok | $0.25 / MTok |

208208| Claude Haiku 5.5 (for prompts over 100,000 tokens) | $0.25 / MTok | $1.25 / MTok |

209209| Claude Haiku 4.5 | $0.50 / MTok | $2.50 / MTok |

from line 247

247247| Claude Opus 4.7 | 675 tokens | 804 tokens |

248248| Claude Opus 4.6 | 497 tokens | 589 tokens |

249249| Claude Opus 4.5 | 496 tokens | 588 tokens |

250| Claude Opus 4.1 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

250| Claude Opus 4.1 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

251251| Claude Opus 4 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

252252| Claude Sonnet 5.5 | 286 tokens | |

253253| Claude Sonnet 5 | 354 tokens | 474 tokens |

254254| Claude Sonnet 4.6 | 497 tokens | 589 tokens |

255255| Claude Sonnet 4.5 | 496 tokens | 588 tokens |

256| Claude Sonnet 4 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

256| Claude Sonnet 4 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

257257| Claude Haiku 5.5 | 286 tokens | 406 tokens |

258258| Claude Haiku 4.5 | 496 tokens | 588 tokens |

259259| Claude Haiku 3.5 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 264 tokens | 355 tokens |

agents-and-tools/agent-skills/quickstart Changed · +7 / -7 lines

from line 34

3434 # List Anthropic-managed Skills

3535 curl --fail-with-body -sS "https://api.anthropic.com/v1/skills?source=anthropic" \

3636 -H "x-api-key: $ANTHROPIC_API_KEY" \

37 -H "anthropic-version: 2023-06-01"

37 -H "anthropic-version: 2026-10-21"

3838 ```

3939

4040 ```bash CLI

from line 130

130130 curl --fail-with-body -sS https://api.anthropic.com/v1/messages \

131131 -H "content-type: application/json" \

132132 -H "x-api-key: $ANTHROPIC_API_KEY" \

133 -H "anthropic-version: 2023-06-01" \

133 -H "anthropic-version: 2026-10-21" \

134134 -d '{

135135 "model": "claude-opus-5-5",

136136 "max_tokens": 16000,

from line 347

347347* **`tools`:** Enables code execution (required for Skills)

348348

349349<Note>

350 The examples use the `code_execution_20260521` tool version, and the Step 3 code parses the result types that current tool versions return. Skills also work with older [code execution tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool) versions such as `code_execution_20250825`: any current code execution tool version satisfies the Skills requirement. If you use a different version, use the tool `type` listed on the code execution tool page.

350 The examples use the `code_execution_20260521` tool version, and the Step 3 code parses the result types that current tool versions return. Skills also work with other [code execution tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool) versions such as `code_execution_20250825`: any current code execution tool version satisfies the Skills requirement. If you use a different version, use the tool `type` listed on the code execution tool page.

351351</Note>

352352

353353When you make this request, Claude automatically matches your task to the relevant Skill. Because you asked for a presentation, Claude determines the PowerPoint Skill is relevant and loads its full instructions: the second level of progressive disclosure. Then Claude runs the Skill’s code to create your presentation.

from line 375

375375 # Download the file and save it

376376 curl --fail-with-body -sS "https://api.anthropic.com/v1/files/$file_id/content" \

377377 -H "x-api-key: $ANTHROPIC_API_KEY" \

378 -H "anthropic-version: 2023-06-01" \

378 -H "anthropic-version: 2026-10-21" \

379379 -o "${TMPDIR:-/tmp}/renewable_energy.pptx"

380380 fi

381381 ```

from line 602

602602 curl --fail-with-body -sS https://api.anthropic.com/v1/messages \

603603 -H "content-type: application/json" \

604604 -H "x-api-key: $ANTHROPIC_API_KEY" \

605 -H "anthropic-version: 2023-06-01" \

605 -H "anthropic-version: 2026-10-21" \

606606 -d '{

607607 "model": "claude-opus-5-5",

608608 "max_tokens": 16000,

from line 794

794794 curl --fail-with-body -sS https://api.anthropic.com/v1/messages \

795795 -H "content-type: application/json" \

796796 -H "x-api-key: $ANTHROPIC_API_KEY" \

797 -H "anthropic-version: 2023-06-01" \

797 -H "anthropic-version: 2026-10-21" \

798798 -d '{

799799 "model": "claude-opus-5-5",

800800 "max_tokens": 16000,

from line 986

986986 curl --fail-with-body -sS https://api.anthropic.com/v1/messages \

987987 -H "content-type: application/json" \

988988 -H "x-api-key: $ANTHROPIC_API_KEY" \

989 -H "anthropic-version: 2023-06-01" \

989 -H "anthropic-version: 2026-10-21" \

990990 -d '{

991991 "model": "claude-opus-5-5",

992992 "max_tokens": 16000,

agents-and-tools/mcp-connector Changed · +3 / -3 lines

from line 82

8282 curl https://api.anthropic.com/v1/messages \

8383 -H "Content-Type: application/json" \

8484 -H "X-API-Key: $ANTHROPIC_API_KEY" \

85 -H "anthropic-version: 2023-06-01" \

85 -H "anthropic-version: 2026-10-21" \

8686 -H "anthropic-beta: mcp-client-2025-11-20" \

8787 -d '{

8888 "model": "claude-opus-5-5",

from line 607

607607 curl https://api.anthropic.com/v1/messages \

608608 -H "content-type: application/json" \

609609 -H "x-api-key: $ANTHROPIC_API_KEY" \

610 -H "anthropic-version: 2023-06-01" \

610 -H "anthropic-version: 2026-10-21" \

611611 -H "anthropic-beta: mcp-client-2026-09-15" \

612612 -d '{

613613 "model": "claude-opus-5-5",

from line 640

640640 curl https://api.anthropic.com/v1/messages \

641641 -H "content-type: application/json" \

642642 -H "x-api-key: $ANTHROPIC_API_KEY" \

643 -H "anthropic-version: 2023-06-01" \

643 -H "anthropic-version: 2026-10-21" \

644644 -H "anthropic-beta: mcp-client-2026-09-15" \

645645 -d '{

646646 "model": "claude-opus-5-5",

agents-and-tools/tool-use/advisor-tool Changed · +41 / -43 lines

from line 77

7777 ```bash cURL

7878 curl https://api.anthropic.com/v1/messages \

7979 -H "x-api-key: $ANTHROPIC_API_KEY" \

80 -H "anthropic-version: 2023-06-01" \

80 -H "anthropic-version: 2026-10-21" \

8181 -H "anthropic-beta: advisor-tool-2026-03-01" \

8282 -H "content-type: application/json" \

8383 -d '{

8484 "model": "claude-sonnet-5",

85 "max_tokens": 4096,

85 "max_tokens": 16000,

8686 "tools": [

8787 {

8888 "type": "advisor_20260301",

from line 92

9292 ],

9393 "messages": [{

9494 "role": "user",

95 "content": "Build a concurrent worker pool in Go with graceful shutdown."

95 "content": "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief."

9696 }]

9797 }'

9898 ```

from line 100

100100 ```bash CLI

101101 ant beta:messages create --beta advisor-tool-2026-03-01 <<'YAML'

102102 model: claude-sonnet-5

103 max_tokens: 4096

103 max_tokens: 16000

104104 tools:

105105 - type: advisor_20260301

106106 name: advisor

from line 107

107107 model: claude-opus-5-5

108108 messages:

109109 - role: user

110 content: Build a concurrent worker pool in Go with graceful shutdown.

110 content: Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.

111111 YAML

112112 ```

113113

from line 116

116116

117117 response = client.beta.messages.create(

118118 model="claude-sonnet-5",

119 max_tokens=4096,

119 max_tokens=16000,

120120 betas=["advisor-tool-2026-03-01"],

121121 tools=[

122122 {

from line 128

128128 messages=[

129129 {

130130 "role": "user",

131 "content": "Build a concurrent worker pool in Go with graceful shutdown.",

131 "content": "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.",

132132 }

133133 ],

134134 )

from line 141

141141

142142 const response = await client.beta.messages.create({

143143 model: "claude-sonnet-5",

144 max_tokens: 4096,

144 max_tokens: 16000,

145145 betas: ["advisor-tool-2026-03-01"],

146146 tools: [

147147 {

from line 153

153153 messages: [

154154 {

155155 role: "user",

156 content: "Build a concurrent worker pool in Go with graceful shutdown."

156 content: "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief."

157157 }

158158 ]

159159 });

from line 170

170170 var parameters = new MessageCreateParams

171171 {

172172 Model = Messages::Model.ClaudeSonnet5,

173 MaxTokens = 4096,

173 MaxTokens = 16000,

174174 Tools = new BetaToolUnion[]

175175 {

176176 new BetaAdvisorTool20260301

from line 182

182182 [

183183 new BetaUserMessageParam

184184 {

185 Content = "Build a concurrent worker pool in Go with graceful shutdown."

185 Content = "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief."

186186 }

187187 ],

188188 Betas = ["advisor-tool-2026-03-01"]

from line 197

197197

198198 response, err := client.Beta.Messages.New(context.TODO(), anthropic.BetaMessageNewParams{

199199 Model: anthropic.ModelClaudeSonnet5,

200 MaxTokens: 4096,

200 MaxTokens: 16000,

201201 Tools: []anthropic.BetaToolUnionParam{

202202 {OfAdvisorTool20260301: &anthropic.BetaAdvisorTool20260301Param{

203203 Model: anthropic.ModelClaudeOpus5_5,

from line 204

204204 }},

205205 },

206206 Messages: []anthropic.BetaMessageParam{

207 anthropic.NewBetaUserMessage(anthropic.NewBetaTextBlock("Build a concurrent worker pool in Go with graceful shutdown.")),

207 anthropic.NewBetaUserMessage(anthropic.NewBetaTextBlock("Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.")),

208208 },

209209 Betas: []anthropic.AnthropicBeta{

210210 anthropic.AnthropicBetaAdvisorTool2026_03_01,

from line 227

227227

228228 MessageCreateParams params = MessageCreateParams.builder()

229229 .model(Model.CLAUDE_SONNET_5)

230 .maxTokens(4096L)

230 .maxTokens(16000L)

231231 .addTool(BetaAdvisorTool20260301.builder()

232232 .model(Model.CLAUDE_OPUS_5_5)

233233 .build())

234 .addUserMessage("Build a concurrent worker pool in Go with graceful shutdown.")

234 .addUserMessage("Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.")

235235 .addBeta("advisor-tool-2026-03-01")

236236 .build();

237237

from line 244

244244 $client = new Client();

245245

246246 $response = $client->beta->messages->create(

247 maxTokens: 4096,

247 maxTokens: 16000,

248248 messages: [

249249 [

250250 'role' => 'user',

251 'content' => 'Build a concurrent worker pool in Go with graceful shutdown.',

251 'content' => 'Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.',

252252 ],

253253 ],

254254 model: Model::CLAUDE_SONNET_5,

from line 270

270270

271271 response = client.beta.messages.create(

272272 model: "claude-sonnet-5",

273 max_tokens: 4096,

273 max_tokens: 16000,

274274 tools: [

275275 {

276276 type: "advisor_20260301",

from line 281

281281 messages: [

282282 {

283283 role: "user",

284 content: "Build a concurrent worker pool in Go with graceful shutdown."

284 content: "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief."

285285 }

286286 ],

287287 betas: ["advisor-tool-2026-03-01"]

from line 454

454454 messages = [

455455 {

456456 "role": "user",

457 "content": "Build a concurrent worker pool in Go with graceful shutdown.",

457 "content": "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.",

458458 }

459459 ]

460460

461461 response = client.beta.messages.create(

462462 model="claude-sonnet-5",

463 max_tokens=1024,

463 max_tokens=16000,

464464 betas=["advisor-tool-2026-03-01"],

465465 tools=tools,

466466 messages=messages,

from line 474

474474

475475 response = client.beta.messages.create(

476476 model="claude-sonnet-5",

477 max_tokens=1024,

477 max_tokens=16000,

478478 betas=["advisor-tool-2026-03-01"],

479479 tools=tools,

480480 messages=messages,

from line 495

495495 const messages: Anthropic.Beta.Messages.BetaMessageParam[] = [

496496 {

497497 role: "user",

498 content: "Build a concurrent worker pool in Go with graceful shutdown."

498 content: "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief."

499499 }

500500 ];

501501

502502 const response = await client.beta.messages.create({

503503 model: "claude-sonnet-5",

504 max_tokens: 1024,

504 max_tokens: 16000,

505505 betas: ["advisor-tool-2026-03-01"],

506506 tools,

507507 messages

from line 515

515515

516516 const followUp = await client.beta.messages.create({

517517 model: "claude-sonnet-5",

518 max_tokens: 1024,

518 max_tokens: 16000,

519519 betas: ["advisor-tool-2026-03-01"],

520520 tools,

521521 messages

from line 526

526526 using Anthropic.Models.Beta.Messages;

527527 using Messages = Anthropic.Models.Messages;

528528

529 // Advisor calls run a second model inside the request, so allow more than the

530 // C# SDK's default non-streaming timeout, which it sizes from max_tokens.

531 var client = new AnthropicClient { Timeout = TimeSpan.FromMinutes(5) };

529 var client = new AnthropicClient();

532530

533531 var tools = new BetaToolUnion[]

534532 {

from line 535

537535

538536 var messages = new List<BetaMessageParam>

539537 {

540 new BetaUserMessageParam { Content = "Build a concurrent worker pool in Go with graceful shutdown." }

538 new BetaUserMessageParam { Content = "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief." }

541539 };

542540

543541 var response = await client.Beta.Messages.Create(new MessageCreateParams

544542 {

545543 Model = Messages::Model.ClaudeSonnet5,

546 MaxTokens = 1024,

544 MaxTokens = 16000,

547545 Tools = tools,

548546 Messages = messages,

549547 Betas = ["advisor-tool-2026-03-01"]

from line 559

561559 var followUp = await client.Beta.Messages.Create(new MessageCreateParams

562560 {

563561 Model = Messages::Model.ClaudeSonnet5,

564 MaxTokens = 1024,

562 MaxTokens = 16000,

565563 Tools = tools,

566564 Messages = messages,

567565 Betas = ["advisor-tool-2026-03-01"]

from line 576

578576 }

579577

580578 messages := []anthropic.BetaMessageParam{

581 anthropic.NewBetaUserMessage(anthropic.NewBetaTextBlock("Build a concurrent worker pool in Go with graceful shutdown.")),

579 anthropic.NewBetaUserMessage(anthropic.NewBetaTextBlock("Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.")),

582580 }

583581

584582 response, err := client.Beta.Messages.New(context.TODO(), anthropic.BetaMessageNewParams{

585583 Model: anthropic.ModelClaudeSonnet5,

586 MaxTokens: 1024,

584 MaxTokens: 16000,

587585 Tools: tools,

588586 Messages: messages,

589587 Betas: []anthropic.AnthropicBeta{

from line 600

602600

603601 response, err = client.Beta.Messages.New(context.TODO(), anthropic.BetaMessageNewParams{

604602 Model: anthropic.ModelClaudeSonnet5,

605 MaxTokens: 1024,

603 MaxTokens: 16000,

606604 Tools: tools,

607605 Messages: messages,

608606 Betas: []anthropic.AnthropicBeta{

from line 631

633631

634632 List<BetaMessageParam> messages = new ArrayList<>();

635633 messages.add(BetaMessageParam.ofUser(

636 "Build a concurrent worker pool in Go with graceful shutdown."));

634 "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief."));

637635

638636 BetaMessage response = client.beta().messages().create(MessageCreateParams.builder()

639637 .model(Model.CLAUDE_SONNET_5)

640 .maxTokens(1024L)

638 .maxTokens(16000L)

641639 .tools(tools)

642640 .messages(messages)

643641 .addBeta("advisor-tool-2026-03-01")

from line 649

651649

652650 BetaMessage followUp = client.beta().messages().create(MessageCreateParams.builder()

653651 .model(Model.CLAUDE_SONNET_5)

654 .maxTokens(1024L)

652 .maxTokens(16000L)

655653 .tools(tools)

656654 .messages(messages)

657655 .addBeta("advisor-tool-2026-03-01")

from line 671

673671 $messages = [

674672 [

675673 'role' => 'user',

676 'content' => 'Build a concurrent worker pool in Go with graceful shutdown.',

674 'content' => 'Build a concurrent worker pool in Go with graceful shutdown. Keep it brief.',

677675 ],

678676 ];

679677

680678 $response = $client->beta->messages->create(

681 maxTokens: 1024,

679 maxTokens: 16000,

682680 messages: $messages,

683681 model: 'claude-sonnet-5',

684682 tools: $tools,

from line 690

692690 $messages[] = ['role' => 'user', 'content' => 'Now add a max-in-flight limit of 10.'];

693691

694692 $response = $client->beta->messages->create(

695 maxTokens: 1024,

693 maxTokens: 16000,

696694 messages: $messages,

697695 model: 'claude-sonnet-5',

698696 tools: $tools,

from line 712

714712 messages = [

715713 {

716714 role: "user",

717 content: "Build a concurrent worker pool in Go with graceful shutdown."

715 content: "Build a concurrent worker pool in Go with graceful shutdown. Keep it brief."

718716 }

719717 ]

720718

721719 response = client.beta.messages.create(

722720 model: "claude-sonnet-5",

723 max_tokens: 1024,

721 max_tokens: 16000,

724722 tools: tools,

725723 messages: messages,

726724 betas: ["advisor-tool-2026-03-01"]

from line 732

734732

735733 response = client.beta.messages.create(

736734 model: "claude-sonnet-5",

737 max_tokens: 1024,

735 max_tokens: 16000,

738736 tools: tools,

739737 messages: messages,

740738 betas: ["advisor-tool-2026-03-01"]

agents-and-tools/tool-use/bash-tool Changed · +3 / -3 lines

from line 54

5454 curl https://api.anthropic.com/v1/messages \

5555 -H "content-type: application/json" \

5656 -H "x-api-key: $ANTHROPIC_API_KEY" \

57 -H "anthropic-version: 2023-06-01" \

57 -H "anthropic-version: 2026-10-21" \

5858 -d '{

5959 "model": "claude-opus-5-5",

6060 "max_tokens": 1024,

from line 280

280280

281281`bash_20250124` is the current version of the tool, and it requires no beta header. Every model from Claude Sonnet 3.7 ([retired](https://platform.claude.com/docs/en/about-claude/model-deprecations)) onward accepts it, including all current Claude models.

282282

283The original `bash_20241022` version works only with the October 2024 Claude Sonnet 3.5 model ([retired](https://platform.claude.com/docs/en/about-claude/model-deprecations)). Requests that use it need the `anthropic-beta: computer-use-2024-10-22` header, and the SDK exposes it only in its beta namespace. New integrations should use `bash_20250124`.

283The `bash_20241022` version works only with the October 2024 Claude Sonnet 3.5 model ([retired](https://platform.claude.com/docs/en/about-claude/model-deprecations)). Requests that use it need the `anthropic-beta: computer-use-2024-10-22` header, and the SDK exposes it only in its beta namespace. New integrations should use `bash_20250124`.

284284

285285## Example: Multistep automation

286286

from line 862

862862 curl https://api.anthropic.com/v1/messages \

863863 -H "content-type: application/json" \

864864 -H "x-api-key: $ANTHROPIC_API_KEY" \

865 -H "anthropic-version: 2023-06-01" \

865 -H "anthropic-version: 2026-10-21" \

866866 -d '{

867867 "model": "claude-opus-5-5",

868868 "max_tokens": 1024,

agents-and-tools/tool-use/browser-use-tool Changed · +11 / -11 lines

from line 48

4848 curl https://api.anthropic.com/v1/messages \

4949 -H "content-type: application/json" \

5050 -H "x-api-key: $ANTHROPIC_API_KEY" \

51 -H "anthropic-version: 2023-06-01" \

51 -H "anthropic-version: 2026-10-21" \

5252 -d '{

5353 "model": "claude-opus-5-5",

5454 "max_tokens": 2048,

from line 1552

15521552

15531553The API validates the toolset entry and every member `tool_use` and `tool_result` block in the conversation. When one is malformed, the API returns an `invalid_request_error` before Claude runs. In the following table, the left column names what you sent.

15541554

1555| Request | Why it fails and what to do |

1556| -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

1557| An option or combination the toolset entry doesn’t accept, for example, a `name`, `strict: true`, `input_examples`, `defer_loading` on the entry itself, a `configs` key that isn’t a member name, a field other than `enabled` or `defer_loading` in a member’s `configs` value ([Configure the toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#configure-the-toolset)), enabled members whose `defer_loading` values differ ([Configure the toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#configure-the-toolset)), a `configs` that leaves no member enabled, a code execution caller in `allowed_callers`, the legacy `fine-grained-tool-streaming-2025-05-14` beta header on the request, a `tool_choice` of type `tool` naming `browser` or a member, or a second browser toolset entry or another tool named `browser` | These aren’t supported on client toolsets. See [Client toolsets](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets) for each rule and its alternative. |

1558| A `tool_result` answering a member call without `"toolset_name": "browser"` or with a different value, or `toolset_name` on a result whose call wasn’t a member call | Echo `toolset_name` exactly on member results, and only on them. |

1559| A member `tool_use` from an earlier turn with no matching `tool_result` | Answer every member call, including the ones you didn’t run after a failure. |

1560| A content block other than `text`, `image`, or `browser_state` in a member result | Member results accept only those three block types. |

1561| A `browser_state` block that breaks a rule in [Track tabs with `browser_state`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#track-tabs-and-page-state), for example, one on an `is_error: true` result or on a result that doesn’t answer a browser member call, more than one in a result, a non-empty `tabs` without exactly one `active: true` entry, a duplicate `tab_id`, an empty `state_changes` array, a `tab_opened` whose `tab_id` isn’t in `tabs`, two state changes for one `download_id` or a state-change field its `type` doesn’t declare ([Report downloads](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#report-downloads)), or a field over its limits | Fix the block. “Nothing to report” is expressed by omitting the block or the `state_changes` field, never by an empty value. |

1562| A successful `new_tab`, `switch_tab`, `close_tab`, or `list_tabs` result whose `content` isn’t exactly one `browser_state` block, or a `new_tab` result without exactly one `tab_opened` matching the active tab | The API renders these results from the block and needs it in that exact shape; see [Tab management results](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#tab-management-results). |

1563| An `image` in a result over your model’s [image size limits](https://platform.claude.com/docs/en/build-with-claude/vision#evaluate-image-size), or over the stricter per-image limit that applies once the request holds [more than 20 images](https://platform.claude.com/docs/en/build-with-claude/vision#request-limits), counting screenshots and `zoom` images in earlier results | The API doesn’t downscale toolset images. Resize screenshots before returning them ([Size screenshots to fit image limits](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#handle-coordinate-scaling-for-higher-resolutions)). |

1564| A `model` that doesn’t support `browser_toolset_20260801` | See [API support](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#availability-feature-support) for the supported models. |

1555| Request | Why it fails and what to do |

1556| ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

1557| An option or combination the toolset entry doesn’t accept, for example, a `name`, `strict: true`, `input_examples`, `defer_loading` on the entry itself, a `configs` key that isn’t a member name, a field other than `enabled` or `defer_loading` in a member’s `configs` value ([Configure the toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#configure-the-toolset)), enabled members whose `defer_loading` values differ ([Configure the toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#configure-the-toolset)), a `configs` that leaves no member enabled, a code execution caller in `allowed_callers`, the `fine-grained-tool-streaming-2025-05-14` beta header on the request, a `tool_choice` of type `tool` naming `browser` or a member, or a second browser toolset entry or another tool named `browser` | These aren’t supported on client toolsets. See [Client toolsets](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets) for each rule and its alternative. |

1558| A `tool_result` answering a member call without `"toolset_name": "browser"` or with a different value, or `toolset_name` on a result whose call wasn’t a member call | Echo `toolset_name` exactly on member results, and only on them. |

1559| A member `tool_use` from an earlier turn with no matching `tool_result` | Answer every member call, including the ones you didn’t run after a failure. |

1560| A content block other than `text`, `image`, or `browser_state` in a member result | Member results accept only those three block types. |

1561| A `browser_state` block that breaks a rule in [Track tabs with `browser_state`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#track-tabs-and-page-state), for example, one on an `is_error: true` result or on a result that doesn’t answer a browser member call, more than one in a result, a non-empty `tabs` without exactly one `active: true` entry, a duplicate `tab_id`, an empty `state_changes` array, a `tab_opened` whose `tab_id` isn’t in `tabs`, two state changes for one `download_id` or a state-change field its `type` doesn’t declare ([Report downloads](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#report-downloads)), or a field over its limits | Fix the block. “Nothing to report” is expressed by omitting the block or the `state_changes` field, never by an empty value. |

1562| A successful `new_tab`, `switch_tab`, `close_tab`, or `list_tabs` result whose `content` isn’t exactly one `browser_state` block, or a `new_tab` result without exactly one `tab_opened` matching the active tab | The API renders these results from the block and needs it in that exact shape; see [Tab management results](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#tab-management-results). |

1563| An `image` in a result over your model’s [image size limits](https://platform.claude.com/docs/en/build-with-claude/vision#evaluate-image-size), or over the stricter per-image limit that applies once the request holds [more than 20 images](https://platform.claude.com/docs/en/build-with-claude/vision#request-limits), counting screenshots and `zoom` images in earlier results | The API doesn’t downscale toolset images. Resize screenshots before returning them ([Size screenshots to fit image limits](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#handle-coordinate-scaling-for-higher-resolutions)). |

1564| A `model` that doesn’t support `browser_toolset_20260801` | See [API support](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool#availability-feature-support) for the supported models. |

15651565

15661566## Limitations

15671567

agents-and-tools/tool-use/code-execution-tool Changed · +8 / -6 lines

from line 63

6363To check whether a model accepts the code execution tool before you send a request, read its `capabilities.server_tools.code_execution.supported` value from the Models API. Don’t use the top-level `capabilities.code_execution` value for this. It reports whether code running in this tool can call your request’s other tools, as in programmatic tool calling. [Using the Models API](https://platform.claude.com/docs/en/models/overview#using-the-models-api) describes both fields.

6464

6565<Note>

66 If you’re still using the legacy `code_execution_20250522` (Python only), see [Upgrade to latest tool version](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool#upgrade-to-latest-tool-version) to migrate from it.

66 If you’re using `code_execution_20250522` (Python only), see [Upgrade to latest tool version](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool#upgrade-to-latest-tool-version) to migrate from it.

6767</Note>

6868

6969## Quick start

from line 74

7474 ```bash cURL

7575 curl --fail-with-body -sS https://api.anthropic.com/v1/messages \

7676 -H "x-api-key: $ANTHROPIC_API_KEY" \

77 -H "anthropic-version: 2023-06-01" \

77 -H "anthropic-version: 2026-10-21" \

7878 -H "content-type: application/json" \

7979 -d '{

8080 "model": "claude-opus-5-5",

from line 295

295295 # First, upload a file

296296 curl --fail-with-body -sS https://api.anthropic.com/v1/files \

297297 -H "x-api-key: $ANTHROPIC_API_KEY" \

298 -H "anthropic-version: 2023-06-01" \

298 -H "anthropic-version: 2026-10-21" \

299299 -F "[email protected]"

300300

301301 # Then use the file with code execution: FILE_ID is the ID from the upload response

302302 curl --fail-with-body -sS https://api.anthropic.com/v1/messages \

303303 -H "x-api-key: $ANTHROPIC_API_KEY" \

304 -H "anthropic-version: 2023-06-01" \

304 -H "anthropic-version: 2026-10-21" \

305305 -H "content-type: application/json" \

306306 -d '{

307307 "model": "claude-opus-5-5",

from line 1091

10911091

10921092The response might include a `pause_turn` stop reason, which indicates that the API paused a long-running turn. You may provide the response back as-is in a subsequent request to let Claude continue its turn, or modify the content if you want to interrupt the conversation.

10931093

1094Send the response’s container ID back with that request (see [Container reuse](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool#container-reuse)). Without it, the request runs in a new container, and files from earlier in the turn are lost.

1095

10941096## Containers

10951097

10961098The code execution tool runs in a secure, containerized environment designed specifically for code execution, with a higher focus on Python.

from line 1146

11441146 # First request: Create a file with a random number

11451147 curl https://api.anthropic.com/v1/messages \

11461148 -H "x-api-key: $ANTHROPIC_API_KEY" \

1147 -H "anthropic-version: 2023-06-01" \

1149 -H "anthropic-version: 2026-10-21" \

11481150 -H "content-type: application/json" \

11491151 -d '{

11501152 "model": "claude-opus-5-5",

from line 1165

11631165 # CONTAINER_ID is the container.id from the first response

11641166 curl https://api.anthropic.com/v1/messages \

11651167 -H "x-api-key: $ANTHROPIC_API_KEY" \

1166 -H "anthropic-version: 2023-06-01" \

1168 -H "anthropic-version: 2026-10-21" \

11671169 -H "content-type: application/json" \

11681170 -d '{

11691171 "container": "'"$CONTAINER_ID"'",

agents-and-tools/tool-use/computer-use-tool Changed · +5 / -5 lines

from line 34

3434 Microsoft Foundry: stable

3535 details:

3636 - Claude 5.5 and later models support computer use through the `computer_toolset_20260801` toolset. They return a 400 error for any computer use tool version they don't accept, such as `computer_20251124` on the Claude API. To move a `computer_20251124` integration to the toolset, see [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124). To move from `computer_20250124`, see [Move computer use to the toolset](https://platform.claude.com/docs/en/models/haiku-5-5/migration-guide#computer-use-toolset) in the Claude Haiku 5.5 migration guide.

37 - Claude Opus 4.7, Claude Opus 4.6, Claude Sonnet 4.6, and Claude Opus 4.5 support computer use only through the earlier `computer_20251124` tool version, which requires a beta header; see [Earlier tool versions](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#earlier-tool-versions).

37 - Claude Opus 4.7, Claude Opus 4.6, Claude Sonnet 4.6, and Claude Opus 4.5 support computer use only through the `computer_20251124` tool version, which requires a beta header; see [Earlier tool versions](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#earlier-tool-versions).

3838---

3939

4040<Note>

from line 79

7979 curl https://api.anthropic.com/v1/messages \

8080 -H "content-type: application/json" \

8181 -H "x-api-key: $ANTHROPIC_API_KEY" \

82 -H "anthropic-version: 2023-06-01" \

82 -H "anthropic-version: 2026-10-21" \

8383 -d '{

8484 "model": "claude-opus-5-5",

8585 "max_tokens": 1024,

from line 879

879879To combine computer use with thinking, see [Thinking](https://platform.claude.com/docs/en/build-with-claude/thinking).

880880

881881<Tip>

882 For the earlier `computer_20251124` tool, internal benchmarking on the models that use it suggests these `effort` settings:

882 For the `computer_20251124` tool, internal benchmarking on the models that use it suggests these `effort` settings:

883883

884884 * **Claude Opus 4.7:** use `high` as the default; use `low` for high-throughput or cost-sensitive workloads.

885885 * **Claude Sonnet 4.6 and Claude Opus 4.6:** use `medium` as the default (best accuracy-to-cost ratio). Avoid `max`, which adds token cost without improving accuracy on UI tasks. On these models, `low` uses *fewer* output tokens than disabling thinking entirely (fewer mistakes mean fewer retries), making it a strong option for cost-sensitive loops.

from line 1842

18421842| Accuracy is consistently poor | Resolution too low | Try 1280x720 as a baseline |

18431843

18441844<Tip>

1845 **Model choice affects click precision.** Among the models that use the earlier `computer_20251124` tool, Claude Sonnet 4.6 is more mechanically precise at clicking than Claude Opus 4.6 and is more robust when screenshots require heavy downscaling. Claude Opus 4.7 narrows that gap: its click precision is roughly comparable to Sonnet 4.6, and its higher resolution limit means less downscaling is needed.

1845 **Model choice affects click precision.** Among the models that use the `computer_20251124` tool, Claude Sonnet 4.6 is more mechanically precise at clicking than Claude Opus 4.6 and is more robust when screenshots require heavy downscaling. Claude Opus 4.7 narrows that gap: its click precision is roughly comparable to Sonnet 4.6, and its higher resolution limit means less downscaling is needed.

18461846</Tip>

18471847

18481848### Follow implementation best practices

from line 2180

21812181| ------------------- | ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |

21822182| `computer_20251124` | `computer-use-2025-11-24` | Claude Fable 5.1, Claude Mythos 5.1, Claude Fable 5, Claude Mythos 5, Claude Opus 5, Claude Sonnet 5, Claude Opus 4.8, Claude Opus 4.7, Claude Opus 4.6, Claude Sonnet 4.6, and Claude Opus 4.5 | [API reference](https://platform.claude.com/docs/en/api/beta/messages/create) |

2183| `computer_20250124` | `computer-use-2025-01-24` | Claude Sonnet 4.5 ([deprecated](https://platform.claude.com/docs/en/about-claude/model-deprecations)), Claude Haiku 4.5, Claude Opus 4.1 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)), Claude Sonnet 4 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)), and Claude Opus 4 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | [API reference](https://platform.claude.com/docs/en/api/beta/messages/create) |

2183| `computer_20250124` | `computer-use-2025-01-24` | Claude Sonnet 4.5 ([deprecated](https://platform.claude.com/docs/en/about-claude/model-deprecations)), Claude Haiku 4.5, Claude Opus 4.1 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)), Claude Sonnet 4 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)), and Claude Opus 4 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | [API reference](https://platform.claude.com/docs/en/api/beta/messages/create) |

21842184

21852185On Amazon Bedrock, Claude Opus 5.5, Claude Sonnet 5.5, and Claude Haiku 5.5 also accept the `computer_20251124` tool with the `computer-use-2025-11-24` beta header.

21862186

agents-and-tools/tool-use/overview Changed · +3 / -3 lines

from line 12

1212 ```bash cURL

1313 curl https://api.anthropic.com/v1/messages \

1414 -H "x-api-key: $ANTHROPIC_API_KEY" \

15 -H "anthropic-version: 2023-06-01" \

15 -H "anthropic-version: 2026-10-21" \

1616 -H "content-type: application/json" \

1717 -d '{

1818 "model": "claude-opus-5-5",

from line 770

770770| Claude Opus 4.7 | 675 tokens | 804 tokens |

771771| Claude Opus 4.6 | 497 tokens | 589 tokens |

772772| Claude Opus 4.5 | 496 tokens | 588 tokens |

773| Claude Opus 4.1 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

773| Claude Opus 4.1 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

774774| Claude Opus 4 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

775775| Claude Sonnet 5.5 | 286 tokens | |

776776| Claude Sonnet 5 | 354 tokens | 474 tokens |

777777| Claude Sonnet 4.6 | 497 tokens | 589 tokens |

778778| Claude Sonnet 4.5 | 496 tokens | 588 tokens |

779| Claude Sonnet 4 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

779| Claude Sonnet 4 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 313 tokens | 315 tokens |

780780| Claude Haiku 5.5 | 286 tokens | 406 tokens |

781781| Claude Haiku 4.5 | 496 tokens | 588 tokens |

782782| Claude Haiku 3.5 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | 264 tokens | 355 tokens |

agents-and-tools/tool-use/programmatic-tool-calling Changed · +18 / -16 lines

from line 50

5050 ```bash cURL

5151 curl https://api.anthropic.com/v1/messages \

5252 -H "x-api-key: $ANTHROPIC_API_KEY" \

53 -H "anthropic-version: 2023-06-01" \

53 -H "anthropic-version: 2026-10-21" \

5454 -H "content-type: application/json" \

5555 -d '{

5656 "model": "claude-opus-5-5",

from line 528

528528 ```bash cURL

529529 curl https://api.anthropic.com/v1/messages \

530530 -H "x-api-key: $ANTHROPIC_API_KEY" \

531 -H "anthropic-version: 2023-06-01" \

531 -H "anthropic-version: 2026-10-21" \

532532 -H "content-type: application/json" \

533533 -d '{

534534 "model": "claude-opus-5-5",

from line 810

810810 },

811811 new AssistantMessageParam

812812 {

813 Content = new ContentBlock[]

813 Content = new List<ContentBlockParam>

814814 {

815 new TextBlock { Text = "I'll query the purchase history and analyze the results." },

816 new ServerToolUseBlock

815 new TextBlockParam { Text = "I'll query the purchase history and analyze the results." },

816 new ServerToolUseBlockParam

817817 {

818 Id = "srvtoolu_abc123",

819 Name = "code_execution",

820 Input = new { code = "..." }

818 ID = "srvtoolu_abc123",

819 Name = ServerToolUseBlockParamName.CodeExecution,

820 Input = new Dictionary<string, JsonElement>

821 {

822 ["code"] = JsonSerializer.SerializeToElement("..."),

823 }

821824 },

822 new ToolUseBlock

825 new ToolUseBlockParam

823826 {

824 Id = "toolu_def456",

827 ID = "toolu_def456",

825828 Name = "query_database",

826 Input = new { sql = "<sql>" },

827 Caller = new ToolCaller

829 Input = new Dictionary<string, JsonElement>

828830 {

829 Type = "code_execution_20260120",

830 ToolId = "srvtoolu_abc123"

831 }

831 ["sql"] = JsonSerializer.SerializeToElement("<sql>"),

832 },

833 Caller = new ServerToolCaller20260120 { ToolID = "srvtoolu_abc123" }

832834 }

833835 }

834836 },

835837 new UserMessageParam

836838 {

837 Content = new ContentBlockParam[]

839 Content = new List<ContentBlockParam>

838840 {

839841 new ToolResultBlockParam

840842 {

agents-and-tools/tool-use/server-tools Changed · +4 / -3 lines

from line 33

3333 # the turn by re-sending the request with the assistant content appended to messages.

3434 curl https://api.anthropic.com/v1/messages \

3535 -H "x-api-key: $ANTHROPIC_API_KEY" \

36 -H "anthropic-version: 2023-06-01" \

36 -H "anthropic-version: 2026-10-21" \

3737 -H "content-type: application/json" \

3838 -d '{

3939 "model": "claude-opus-5-5",

from line 392

392392

393393* **Continue the conversation:** Pass the paused response back as-is in a subsequent request to let Claude continue its turn.

394394* **Preserve tool state:** Include the same tools in the continuation request. A paused turn can end with a `server_tool_use` block whose tool has not run yet, and the API returns a validation error if that tool is missing from the continuation.

395* **Repeat as needed:** A continued turn can pause again. Check `stop_reason` on each response and continue until you get a different stop reason, capping the number of continuations as you would any retry loop.

395* **Keep the container:** If the response includes a `container`, as it does with the [code execution tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool), send its ID back in the continuation request. Without it, the request runs in a new container, and files from earlier in the turn are lost.

396* **Repeat as needed:** A continued turn can pause again. Check `stop_reason` on each response and continue until you get a different stop reason, capping the number of continuations as you would any retry loop. Keep every paused response in `messages`, in order. A continuation holds only new content, so a request that leaves out an earlier paused response loses that work and can return a 400 error.

396397

397398For the other `stop_reason` values and general handling patterns, see [Stop reasons and fallback](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons).

398399

from line 508

507508 # and the same tools array (see the SDK tabs).

508509 curl https://api.anthropic.com/v1/messages \

509510 -H "x-api-key: $ANTHROPIC_API_KEY" \

510 -H "anthropic-version: 2023-06-01" \

511 -H "anthropic-version: 2026-10-21" \

511512 -H "content-type: application/json" \

512513 -d '{

513514 "model": "claude-opus-5-5",

agents-and-tools/tool-use/strict-tool-use Changed · +3 / -3 lines

from line 62

6262 curl https://api.anthropic.com/v1/messages \

6363 -H "content-type: application/json" \

6464 -H "x-api-key: $ANTHROPIC_API_KEY" \

65 -H "anthropic-version: 2023-06-01" \

65 -H "anthropic-version: 2026-10-21" \

6666 -d '{

6767 "model": "claude-opus-5-5",

6868 "max_tokens": 1024,

from line 423

423423 curl https://api.anthropic.com/v1/messages \

424424 -H "content-type: application/json" \

425425 -H "x-api-key: $ANTHROPIC_API_KEY" \

426 -H "anthropic-version: 2023-06-01" \

426 -H "anthropic-version: 2026-10-21" \

427427 -d '{

428428 "model": "claude-opus-5-5",

429429 "max_tokens": 1024,

from line 724

724724 curl https://api.anthropic.com/v1/messages \

725725 -H "content-type: application/json" \

726726 -H "x-api-key: $ANTHROPIC_API_KEY" \

727 -H "anthropic-version: 2023-06-01" \

727 -H "anthropic-version: 2026-10-21" \

728728 -d '{

729729 "model": "claude-opus-5-5",

730730 "max_tokens": 1024,

agents-and-tools/tool-use/text-editor-tool Changed · +4 / -4 lines

from line 62

6262 curl https://api.anthropic.com/v1/messages \

6363 -H "content-type: application/json" \

6464 -H "x-api-key: $ANTHROPIC_API_KEY" \

65 -H "anthropic-version: 2023-06-01" \

65 -H "anthropic-version: 2026-10-21" \

6666 -d '{

6767 "model": "claude-opus-5-5",

6868 "max_tokens": 1024,

from line 417

417417 curl https://api.anthropic.com/v1/messages \

418418 -H "content-type: application/json" \

419419 -H "x-api-key: $ANTHROPIC_API_KEY" \

420 -H "anthropic-version: 2023-06-01" \

420 -H "anthropic-version: 2026-10-21" \

421421 -d '{

422422 "model": "claude-opus-5-5",

423423 "max_tokens": 1024,

from line 617

617617 curl https://api.anthropic.com/v1/messages \

618618 -H "content-type: application/json" \

619619 -H "x-api-key: $ANTHROPIC_API_KEY" \

620 -H "anthropic-version: 2023-06-01" \

620 -H "anthropic-version: 2026-10-21" \

621621 -d '{

622622 "model": "claude-opus-5-5",

623623 "max_tokens": 1024,

from line 1080

10801080 curl https://api.anthropic.com/v1/messages \

10811081 -H "content-type: application/json" \

10821082 -H "x-api-key: $ANTHROPIC_API_KEY" \

1083 -H "anthropic-version: 2023-06-01" \

1083 -H "anthropic-version: 2026-10-21" \

10841084 -d '{

10851085 "model": "claude-opus-5-5",

10861086 "max_tokens": 1024,

agents-and-tools/tool-use/web-fetch-tool Changed · +4 / -4 lines

from line 41

4141

4242The web fetch tool allows Claude to retrieve full content from specified webpages and PDF documents.

4343

44The latest web fetch tool version (`web_fetch_20260318`) supports **dynamic filtering**: Claude can write and execute code to filter fetched content before it reaches the context window, keeping only relevant information and discarding the rest. This reduces token consumption while maintaining response quality. Dynamic filtering is available with Claude 4.6 and later models and [Claude Mythos Preview](https://anthropic.com/glasswing). `web_fetch_20260318` also adds [response inclusion](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool#response-inclusion) control for agentic workflows. The previous versions (`web_fetch_20260309` for dynamic filtering and [cache bypass](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool#cache-bypass), `web_fetch_20260209` for dynamic filtering only, `web_fetch_20250910` for basic fetch) remain available.

44The latest web fetch tool version (`web_fetch_20260318`) supports **dynamic filtering**: Claude can write and run code to filter fetched content before it reaches the context window, keeping only relevant information and discarding the rest. This reduces token consumption while maintaining response quality. Dynamic filtering is available with Claude 4.6 and later models and [Claude Mythos Preview](https://anthropic.com/glasswing). `web_fetch_20260318` also adds [response inclusion](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool#response-inclusion) control for agentic workflows. The other versions (`web_fetch_20260309` for dynamic filtering and [cache bypass](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool#cache-bypass), `web_fetch_20260209` for dynamic filtering only, `web_fetch_20250910` for basic fetch) are available.

4545

4646Web fetch (with and without dynamic filtering) is available on the Claude API, [Claude Platform on AWS](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws), [Claude Models on Google Cloud (Anthropic managed)](https://platform.claude.com/docs/en/build-with-claude/claude-models-on-google-cloud/anthropic-managed), and [Microsoft Foundry](https://platform.claude.com/docs/en/build-with-claude/claude-in-microsoft-foundry). On Microsoft Foundry, deployments [hosted on Azure](https://platform.claude.com/docs/en/build-with-claude/claude-in-microsoft-foundry#additional-features-not-supported-when-hosted-on-azure) support only the basic web fetch tool (`web_fetch_20250910`, without dynamic filtering). Deployments hosted on Anthropic support all versions.

4747

from line 117

117117 ```bash cURL

118118 curl https://api.anthropic.com/v1/messages \

119119 -H "x-api-key: $ANTHROPIC_API_KEY" \

120 -H "anthropic-version: 2023-06-01" \

120 -H "anthropic-version: 2026-10-21" \

121121 -H "content-type: application/json" \

122122 -d '{

123123 "model": "claude-opus-5-5",

from line 287

287287 ```bash cURL

288288 curl https://api.anthropic.com/v1/messages \

289289 -H "x-api-key: $ANTHROPIC_API_KEY" \

290 -H "anthropic-version: 2023-06-01" \

290 -H "anthropic-version: 2026-10-21" \

291291 -H "content-type: application/json" \

292292 -d '{

293293 "model": "claude-opus-5-5",

from line 752

752752 ```bash cURL

753753 curl https://api.anthropic.com/v1/messages \

754754 -H "x-api-key: $ANTHROPIC_API_KEY" \

755 -H "anthropic-version: 2023-06-01" \

755 -H "anthropic-version: 2026-10-21" \

756756 -H "content-type: application/json" \

757757 -d '{

758758 "model": "claude-opus-5-5",

api/claude-platform-on-aws-iam-actions Changed · +29 / -24 lines

from line 81

8181</Note>

8282

8383<Note>

84 Creating or deleting an individual skill version maps to `UpdateSkill`, not `CreateSkill` or `DeleteSkill`. A policy that denies `aws-external-anthropic:Delete*` still allows version deletion, and a policy that denies `aws-external-anthropic:Create*` still allows version creation. Deny `UpdateSkill` and `CreateSkill` as well if you need to prevent any skill mutation.

84 Creating or deleting an individual skill version maps to `UpdateSkill`, not `CreateSkill` or `DeleteSkill`. A policy that denies `aws-external-anthropic:Delete*` still allows version deletion, and a policy that denies `aws-external-anthropic:Create*` still allows version creation. To block every write route under `/v1/skills`, deny `CreateSkill`, `UpdateSkill`, and `DeleteSkill`.

8585</Note>

8686

8787### Agents

from line 95

9595| `ArchiveAgent` | `POST /v1/agents/{id}/archive` |

9696

9797<Note>

98 Agents support only archive, not hard delete. A policy that denies `aws-external-anthropic:Delete*` does not block `ArchiveAgent`. Deny `ArchiveAgent`, `UpdateAgent`, and `CreateAgent` if you need to prevent any agent mutation.

98 Agents support only archive, not hard delete. A policy that denies `aws-external-anthropic:Delete*` does not block `ArchiveAgent`. To block every write route under `/v1/agents`, deny `CreateAgent`, `UpdateAgent`, and `ArchiveAgent`.

9999</Note>

100100

101101### Sessions

102102

103| Action | Routes authorized |

104| ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

105| `CreateSession` | `POST /v1/sessions` |

106| `GetSession` | `GET /v1/sessions/{id}` `GET /v1/sessions/{id}/events` `GET /v1/sessions/{id}/events/stream` `GET /v1/sessions/{id}/resources` `GET /v1/sessions/{id}/resources/{id}` |

107| `ListSessions` | `GET /v1/sessions` |

108| `UpdateSession` | `POST /v1/sessions/{id}` `POST /v1/sessions/{id}/events` `POST /v1/sessions/{id}/resources` `POST /v1/sessions/{id}/resources/{id}` `DELETE /v1/sessions/{id}/resources/{id}` |

109| `ArchiveSession` | `POST /v1/sessions/{id}/archive` |

110| `DeleteSession` | `DELETE /v1/sessions/{id}` |

103| Action | Routes authorized |

104| ---------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

105| `CreateSession` | `POST /v1/sessions` |

106| `GetSession` | `GET /v1/sessions/{id}` `GET /v1/sessions/{id}/events` `GET /v1/sessions/{id}/events/stream` `GET /v1/sessions/{id}/resources` `GET /v1/sessions/{id}/resources/{id}` `GET /v1/sessions/{id}/threads` `GET /v1/sessions/{id}/threads/{id}` `GET /v1/sessions/{id}/threads/{id}/events` `GET /v1/sessions/{id}/threads/{id}/stream` |

107| `ListSessions` | `GET /v1/sessions` |

108| `UpdateSession` | `POST /v1/sessions/{id}` `POST /v1/sessions/{id}/events` `POST /v1/sessions/{id}/resources` `POST /v1/sessions/{id}/resources/{id}` `DELETE /v1/sessions/{id}/resources/{id}` `POST /v1/sessions/{id}/threads/{id}/archive` |

109| `ArchiveSession` | `POST /v1/sessions/{id}/archive` |

110| `DeleteSession` | `DELETE /v1/sessions/{id}` |

111111

112112<Note>

113 `GetSession` authorizes reading session metadata, the full event stream (conversation history), and session resources. The `AnthropicReadOnlyAccess`, `AnthropicInferenceAccess`, and `AnthropicLimitedAccess` policies’ `Get*` wildcards include this action.

113 `GetSession` authorizes reading session metadata, the event stream (conversation history), session resources, and [session threads](https://platform.claude.com/docs/en/managed-agents/session-threads), including each thread’s events. The `AnthropicReadOnlyAccess`, `AnthropicInferenceAccess`, and `AnthropicLimitedAccess` policies’ `Get*` wildcards include this action, and the `AnthropicSelfHostedEnvironmentAccess` policy grants it explicitly.

114114</Note>

115115

116116<Note>

117 Creating, updating, or deleting an individual session sub-resource (events or session resources) maps to `UpdateSession`, not `CreateSession` or `DeleteSession`. A policy that denies `aws-external-anthropic:Delete*` still allows sub-resource deletion, and a policy that denies `aws-external-anthropic:Create*` still allows sub-resource creation. Deny `UpdateSession`, `CreateSession`, and `ArchiveSession` as well if you need to prevent any session mutation.

117 Every route that writes to an existing session’s events, resources, or threads maps to `UpdateSession`, not `CreateSession`, `DeleteSession`, or `ArchiveSession`. A policy that denies `aws-external-anthropic:Create*`, `Delete*`, or `Archive*` still allows sending events, adding or deleting session resources, and archiving session threads. To block every write route under `/v1/sessions`, deny `CreateSession`, `UpdateSession`, `ArchiveSession`, and `DeleteSession`. A principal with `ProcessEnvironmentWork` can still [stop a session](https://platform.claude.com/docs/en/managed-agents/self-hosted-sandboxes-operations#stop-a-session-gracefully) that runs in a self-hosted environment.

118118</Note>

119119

120120### Environments

from line 130

130130| `ProcessEnvironmentWork` | `GET /v1/environments/{id}/work/poll` `POST /v1/environments/{id}/work/{work_id}` `POST /v1/environments/{id}/work/{work_id}/ack` `POST /v1/environments/{id}/work/{work_id}/heartbeat` `POST /v1/environments/{id}/work/{work_id}/stop` |

131131

132132<Note>

133 A policy that denies `aws-external-anthropic:Delete*` does not block `ArchiveEnvironment`. `ProcessEnvironmentWork` is not matched by `Create*`, `Update*`, `Delete*`, or `Archive*` wildcards. Deny `ArchiveEnvironment`, `UpdateEnvironment`, `CreateEnvironment`, and `ProcessEnvironmentWork` as well if you need to prevent any environment mutation.

133 A policy that denies `aws-external-anthropic:Delete*` does not block `ArchiveEnvironment`. `ProcessEnvironmentWork` is not matched by `Create*`, `Update*`, `Delete*`, or `Archive*` wildcards. To block every write route under `/v1/environments`, deny `CreateEnvironment`, `UpdateEnvironment`, `ArchiveEnvironment`, `DeleteEnvironment`, and `ProcessEnvironmentWork`.

134134</Note>

135135

136136<Note>

from line 149

149149| `DeleteVault` | `DELETE /v1/vaults/{id}` |

150150

151151<Note>

152 Creating, updating, archiving, or deleting an individual vault credential maps to `UpdateVault`. Reading a credential maps to `GetVault`. Vault credential secrets are not exposed: secret fields are write-only and are never returned by `GetVault` (see [Authenticate with vaults](https://platform.claude.com/docs/en/managed-agents/vaults)). A policy that denies `aws-external-anthropic:Delete*` still allows credential deletion, and a policy that denies `aws-external-anthropic:Create*` still allows credential creation. Deny `UpdateVault`, `CreateVault`, and `ArchiveVault` as well if you need to prevent any vault mutation.

152 Creating, updating, archiving, or deleting an individual vault credential maps to `UpdateVault`. Reading a credential maps to `GetVault`. Vault credential secrets are not exposed: secret fields are write-only and are never returned by `GetVault` (see [Authenticate with vaults](https://platform.claude.com/docs/en/managed-agents/vaults)). A policy that denies `aws-external-anthropic:Create*`, `Delete*`, or `Archive*` still allows creating, deleting, and archiving credentials. To block every write route under `/v1/vaults`, deny `CreateVault`, `UpdateVault`, `ArchiveVault`, and `DeleteVault`.

153153</Note>

154154

155155### Memory stores

from line 168

168168</Note>

169169

170170<Note>

171 Creating, updating, or deleting an individual memory and redacting a memory version both map to `UpdateMemoryStore`, not `CreateMemoryStore` or `DeleteMemoryStore`. A policy that denies `aws-external-anthropic:Delete*` still allows individual-memory deletion and memory-version redaction, and a policy that denies `aws-external-anthropic:Create*` still allows individual-memory creation. Deny `UpdateMemoryStore`, `CreateMemoryStore`, and `ArchiveMemoryStore` as well if you need to prevent any memory-store mutation.

171 Creating, updating, or deleting an individual memory and redacting a memory version both map to `UpdateMemoryStore`, not `CreateMemoryStore` or `DeleteMemoryStore`. A policy that denies `aws-external-anthropic:Delete*` still allows individual-memory deletion and memory-version redaction, and a policy that denies `aws-external-anthropic:Create*` still allows individual-memory creation. To block every write route under `/v1/memory_stores`, deny `CreateMemoryStore`, `UpdateMemoryStore`, `ArchiveMemoryStore`, and `DeleteMemoryStore`. A principal with `CreateSession` can still [attach a store to a session](https://platform.claude.com/docs/en/managed-agents/memory#attach-a-memory-store-to-a-session) with `read_write` access, the default. That session’s agent can then write to the store.

172172</Note>

173173

174174### Webhooks

from line 187

187187</Note>

188188

189189<Note>

190 `RotateWebhookSecret` is not matched by `aws-external-anthropic:Create*`, `Update*`, or `Delete*` wildcards. A policy that denies those patterns still allows secret rotation. Deny `RotateWebhookSecret`, `UpdateWebhook`, `CreateWebhook`, and `DeleteWebhook` if you need to prevent any webhook mutation.

190 `RotateWebhookSecret` is not matched by `aws-external-anthropic:Create*`, `Update*`, or `Delete*` wildcards. A policy that denies those patterns still allows secret rotation. To block every write route under `/v1/webhooks`, deny `CreateWebhook`, `UpdateWebhook`, `DeleteWebhook`, and `RotateWebhookSecret`.

191191</Note>

192192

193193### User profiles

from line 220

220220| `ArchiveWorkspace` | `POST /v1/organizations/workspaces/{id}/archive` |

221221

222222<Note>

223 Workspaces support only archive, not hard delete. A policy that denies `aws-external-anthropic:Delete*` does not block `ArchiveWorkspace`. Deny `ArchiveWorkspace`, `UpdateWorkspace`, and `CreateWorkspace` if you need to prevent any workspace mutation.

223 Workspaces support only archive, not hard delete. A policy that denies `aws-external-anthropic:Delete*` does not block `ArchiveWorkspace`. To block every write route under `/v1/organizations/workspaces`, deny `CreateWorkspace`, `UpdateWorkspace`, and `ArchiveWorkspace`.

224224</Note>

225225

226226### Encryption keys

from line 234

234234| `DisableKey` | `DELETE /v1/organizations/external_keys/{id}` |

235235

236236<Note>

237 These actions manage your organization’s [customer-managed encryption key (CMEK)](https://platform.claude.com/docs/en/manage-claude/cmek-claude-platform-on-aws) registrations, the record of which AWS KMS key ARNs are registered. They do not create, change, or disable the keys in AWS KMS. `DisableKey` removes a registration and is rejected while any workspace still uses the key. `RegisterKey` and `DisableKey` are not matched by `Create*`, `Update*`, or `Delete*` wildcards; deny `RegisterKey`, `UpdateKey`, and `DisableKey` if you need to prevent any change to key registrations. In these routes, `{id}` is the URL-encoded KMS key ARN. Attaching a registered key to a workspace is a workspace operation, authorized by `CreateWorkspace` or `UpdateWorkspace`; the principal that attaches a key also needs `kms:DescribeKey`, `kms:Encrypt`, and `kms:Decrypt` on that key (see the [prerequisites](https://platform.claude.com/docs/en/manage-claude/cmek-claude-platform-on-aws#prerequisites)). External key actions are account-scoped: specifying a workspace ARN on them has no effect; use `Resource: "*"`.

237 These actions manage your organization’s [customer-managed encryption key (CMEK)](https://platform.claude.com/docs/en/manage-claude/cmek-claude-platform-on-aws) registrations, the record of which AWS KMS key ARNs are registered. They do not create, change, or disable the keys in AWS KMS. `DisableKey` removes a registration and is rejected while any workspace still uses the key. `RegisterKey` and `DisableKey` are not matched by `Create*`, `Update*`, or `Delete*` wildcards. To block every write route under `/v1/organizations/external_keys`, deny `RegisterKey`, `UpdateKey`, and `DisableKey`. In these routes, `{id}` is the URL-encoded KMS key ARN. Attaching a registered key to a workspace is a workspace operation, authorized by `CreateWorkspace` or `UpdateWorkspace`; the principal that attaches a key also needs `kms:DescribeKey`, `kms:Encrypt`, and `kms:Decrypt` on that key (see the [prerequisites](https://platform.claude.com/docs/en/manage-claude/cmek-claude-platform-on-aws#prerequisites)). External key actions are account-scoped: specifying a workspace ARN on them has no effect; use `Resource: "*"`.

238238</Note>

239239

240240### Compliance

from line 270

270270</Note>

271271

272272<Note>

273 `AddTelemetryDestinationWorkspace` and `RemoveTelemetryDestinationWorkspace` apply to the workspace in the route, and only on a destination whose **Workspaces** option is **Selected**. To change an **All except** list, use `UpdateAccountTelemetryDestination`. A policy that denies `aws-external-anthropic:Create*`, `Update*`, or `Delete*` still allows adding and removing workspaces. To block every change to destinations, also deny `AddTelemetryDestinationWorkspace` and `RemoveTelemetryDestinationWorkspace`.

273 `AddTelemetryDestinationWorkspace` and `RemoveTelemetryDestinationWorkspace` apply to the workspace in the route, and only on a destination whose **Workspaces** option is **Selected**. To change an **All except** list, use `UpdateAccountTelemetryDestination`. A policy that denies `aws-external-anthropic:Create*`, `Update*`, or `Delete*` still allows adding and removing workspaces. To block every write route under `/v1/organizations/telemetry_destinations`, deny `CreateAccountTelemetryDestination`, `UpdateAccountTelemetryDestination`, `DeleteAccountTelemetryDestination`, `TestAccountTelemetryDestination`, `AddTelemetryDestinationWorkspace`, and `RemoveTelemetryDestinationWorkspace`.

274274</Note>

275275

276276### Usage and cost

from line 376

376376| `POST` | `/v1/sessions/{id}/resources` | `UpdateSession` | Data |

377377| `POST` | `/v1/sessions/{id}/resources/{id}` | `UpdateSession` | Data |

378378| `DELETE` | `/v1/sessions/{id}/resources/{id}` | `UpdateSession` | Data |

379| `GET` | `/v1/sessions/{id}/threads` | `GetSession` | Data |

380| `GET` | `/v1/sessions/{id}/threads/{id}` | `GetSession` | Data |

381| `GET` | `/v1/sessions/{id}/threads/{id}/events` | `GetSession` | Data |

382| `GET` | `/v1/sessions/{id}/threads/{id}/stream` | `GetSession` | Data |

383| `POST` | `/v1/sessions/{id}/threads/{id}/archive` | `UpdateSession` | Data |

381386| `GET` | `/v1/environments/{id}` | `GetEnvironment` | Data |

api/errors Changed · +6 / -4 lines

from line 12

1212

1313* 401 - `authentication_error`: There’s an issue with your [API key](https://platform.claude.com/docs/en/get-api-key) (for example, it’s malformed, revoked, or expired; see [Key expiration](https://platform.claude.com/docs/en/manage-claude/authentication#key-expiration)). On Claude Platform on AWS, this can also indicate a problem with your AWS credentials or SigV4 signature.

1414

15 A 401 whose message begins `CORS requests are not allowed for this Organization` is not a problem with your API key. The request included an `Origin` header, which browsers add, and your organization cannot call the Claude API directly from a browser. The message continues `because of custom retention settings` or `because of its settings`. The first appears only for data retention arrangements such as zero data retention (ZDR); the second is the general case, which includes the Compliance API. No change to the browser request fixes it, so send the request from your own back end; see [Browser-based applications](https://platform.claude.com/docs/en/manage-claude/authentication#browser-based-applications).

16

1517* 402 - `billing_error`: There’s an issue with your billing or payment information. Check your payment details in the [Claude Console](https://platform.claude.com), or in AWS Marketplace if you’re using Claude Platform on AWS.

1618

1719* 403 - `permission_error`: Your API key does not have permission to use the specified resource. Check your organization’s access and workspace settings in the [Claude Console](https://platform.claude.com).

from line 26

2426

2527* 429 - `rate_limit_error`: Your organization has hit a [rate limit](https://platform.claude.com/docs/en/api/rate-limits), reached its usage tier’s monthly spend cap, or reached a spend limit on the Claude Code workspace. A tier spend-cap 429 has no `retry-after` header and keeps failing until access resumes; see [Reaching your spend cap](https://platform.claude.com/docs/en/api/rate-limits#reaching-your-spend-cap) for how to recognize it.

2628

27* 500 - `api_error`: An unexpected error has occurred internal to Anthropic’s systems. Retry the request with exponential backoff; if the error persists, contact support with the [request ID](https://platform.claude.com/docs/en/api/errors#request-id).

29* 500 - `api_error`: An unexpected error has occurred internal to Anthropic’s systems. Retry the request with exponential backoff; if the error persists, contact support with the [request ID](https://platform.claude.com/docs/en/api/errors#request-id). If the error type is `timeout_error` instead, see the 504 entry.

2830

29* 504 - `timeout_error`: The request timed out while processing. Consider using the [streaming Messages API](https://platform.claude.com/docs/en/build-with-claude/streaming) for long-running requests. See [Long requests](https://platform.claude.com/docs/en/api/errors#long-requests) for more options.

31* 504 - `timeout_error`: The request timed out while processing. The API also returns `timeout_error` with other status codes, such as a 500 when a request reaches the server’s time limit before the response starts. For long-running Messages API requests, use the [streaming Messages API](https://platform.claude.com/docs/en/build-with-claude/streaming). See [Long requests](https://platform.claude.com/docs/en/api/errors#long-requests) for more options.

3032

3133* 529 - `overloaded_error`: The API is temporarily overloaded.

3234

from line 91

8991 # -i prints the response headers, including request-id, before the body

9092 curl -sS -i https://api.anthropic.com/v1/messages \

9193 -H "x-api-key: $ANTHROPIC_API_KEY" \

92 -H "anthropic-version: 2023-06-01" \

94 -H "anthropic-version: 2026-10-21" \

9395 -H "content-type: application/json" \

9496 -d '{

9597 "model": "claude-sonnet-5",

api/migration-2026-10-21 Changed · +7 / -7 lines

from line 140

140140

141141Admin API:

142142

143| Endpoints | Old name | New name |

144| ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- |

145| [RBAC groups](https://platform.claude.com/docs/en/api/beta/organization/rbac_groups/list): create, list, retrieve, and update | `roles` | `role_ids` |

146| [RBAC group members](https://platform.claude.com/docs/en/api/beta/organization/rbac_groups/members/list): list and add, and the response to [removing a member](https://platform.claude.com/docs/en/api/beta/organization/rbac_groups/members/remove) | `group_id` | `rbac_group_id` |

147| [RBAC roles](https://platform.claude.com/docs/en/api/beta/organization/rbac_roles/list): list and retrieve | `name` | `display_name` |

148| [Organization](https://platform.claude.com/docs/en/api/organization/rate_limits/list) and [workspace](https://platform.claude.com/docs/en/api/organization/workspaces/rate_limits/list) rate limits: list, and the response to setting a workspace’s rate limits | `group_type` | `group.type` |

149| [API keys](https://platform.claude.com/docs/en/api/organization/api_keys/list): list, retrieve, and update | `workspace_id` | `scope` (different values; see [`workspace_id` on API keys](https://platform.claude.com/docs/en/api/migration-2026-10-21#api-key-workspace-id)) |

143| Endpoints | Old name | New name |

144| ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- |

145| [RBAC groups](https://platform.claude.com/docs/en/api/beta/organization/rbac_groups/list): create, list, retrieve, and update | `roles` | `role_ids` |

146| [RBAC group members](https://platform.claude.com/docs/en/api/beta/organization/rbac_groups/members/list): list and add, and the response to [removing a member](https://platform.claude.com/docs/en/api/beta/organization/rbac_groups/members/remove) | `group_id` | `rbac_group_id` |

147| [RBAC roles](https://platform.claude.com/docs/en/api/beta/organization/rbac_roles/list): list and retrieve | `name` | `display_name` |

148| [Organization](https://platform.claude.com/docs/en/api/organization/rate_limits/list) and [workspace](https://platform.claude.com/docs/en/api/organization/workspaces/rate_limits/list) rate limits: list, and the response to [setting a workspace’s rate limits](https://platform.claude.com/docs/en/api/organization/workspaces/rate_limits/update) | `group_type` | `group.type` |

149| [API keys](https://platform.claude.com/docs/en/api/organization/api_keys/list): list, retrieve, and update | `workspace_id` | `scope` (different values; see [`workspace_id` on API keys](https://platform.claude.com/docs/en/api/migration-2026-10-21#api-key-workspace-id)) |

150150

151151Claude Enterprise Analytics API:

152152

build-with-claude/batch-processing Changed · +11 / -11 lines

from line 117

117117| Claude Opus 4.7 | $2.50 / MTok | $12.50 / MTok |

118118| Claude Opus 4.6 | $2.50 / MTok | $12.50 / MTok |

119119| Claude Opus 4.5 | $2.50 / MTok | $12.50 / MTok |

120| Claude Opus 4.1 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $7.50 / MTok | $37.50 / MTok |

120| Claude Opus 4.1 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $7.50 / MTok | $37.50 / MTok |

121121| Claude Opus 4 ([retired, except on Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $7.50 / MTok | $37.50 / MTok |

122122| Claude Sonnet 5.5 | $1 / MTok | $5 / MTok |

123123| Claude Sonnet 5 | $1 / MTok | $5 / MTok |

124124| Claude Sonnet 4.6 | $1.50 / MTok | $7.50 / MTok |

125125| Claude Sonnet 4.5 | $1.50 / MTok | $7.50 / MTok |

126| Claude Sonnet 4 ([retired, except on Bedrock and Claude Models on Google Cloud (Google managed)](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $1.50 / MTok | $7.50 / MTok |

126| Claude Sonnet 4 ([retired, except on Claude Models on Google Cloud (Google managed) and Bedrock](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | $1.50 / MTok | $7.50 / MTok |

127127| Claude Haiku 5.5 (for prompts up to 100,000 tokens) | $0.05 / MTok | $0.25 / MTok |

128128| Claude Haiku 5.5 (for prompts over 100,000 tokens) | $0.25 / MTok | $1.25 / MTok |

129129| Claude Haiku 4.5 | $0.50 / MTok | $2.50 / MTok |

from line 148

148148 ```bash cURL

149149 curl https://api.anthropic.com/v1/messages/batches \

150150 -H "x-api-key: $ANTHROPIC_API_KEY" \

151 -H "anthropic-version: 2023-06-01" \

151 -H "anthropic-version: 2026-10-21" \

152152 -H "content-type: application/json" \

153153 -d '{

154154 "requests": [

from line 479

479479 # Check the status; repeat until processing_status is "ended"

480480 curl "https://api.anthropic.com/v1/messages/batches/$MESSAGE_BATCH_ID" \

481481 -H "x-api-key: $ANTHROPIC_API_KEY" \

482 -H "anthropic-version: 2023-06-01"

482 -H "anthropic-version: 2026-10-21"

483483 ```

484484

485485 ```bash CLI

from line 628

628628 # perform automatic pagination.)

629629 curl -s "https://api.anthropic.com/v1/messages/batches?limit=20" \

630630 -H "x-api-key: $ANTHROPIC_API_KEY" \

631 -H "anthropic-version: 2023-06-01"

631 -H "anthropic-version: 2026-10-21"

632632 ```

633633

634634 ```bash CLI

from line 745

745745 # Streams the batch's .jsonl results, one result on each line.

746746 # For per-result handling (retries, validation errors), use the Claude SDK.

747747 curl "https://api.anthropic.com/v1/messages/batches/msgbatch_01HkcTjaV5uDC8jWR4ZsDV8d/results" \

748 -H "anthropic-version: 2023-06-01" \

748 -H "anthropic-version: 2026-10-21" \

749749 -H "x-api-key: $ANTHROPIC_API_KEY"

750750 ```

751751

from line 792

792792 console.log(`Success! ${result.custom_id}`);

793793 break;

794794 case "errored":

795 if (result.result.error.type === "invalid_request_error") {

795 if (result.result.error.error.type === "invalid_request_error") {

796796 // Request body must be fixed before re-sending request

797797 console.log(`Validation error: ${result.custom_id}`);

798798 } else {

from line 931

931931 when Anthropic::Models::Messages::MessageBatchSucceededResult

932932 puts "Success! #{result.custom_id}"

933933 when Anthropic::Models::Messages::MessageBatchErroredResult

934 if outcome.error.type == :invalid_request

934 if outcome.error.error.type == :invalid_request_error

935935 puts "Validation error: #{result.custom_id}"

936936 else

937937 puts "Server error: #{result.custom_id}"

from line 968

968968 ```bash cURL

969969 curl -X POST https://api.anthropic.com/v1/messages/batches/$MESSAGE_BATCH_ID/cancel \

970970 -H "x-api-key: $ANTHROPIC_API_KEY" \

971 -H "anthropic-version: 2023-06-01"

971 -H "anthropic-version: 2026-10-21"

972972 ```

973973

974974 ```bash CLI

from line 1080

10801080 ```bash cURL

10811081 curl https://api.anthropic.com/v1/messages/batches \

10821082 -H "x-api-key: $ANTHROPIC_API_KEY" \

1083 -H "anthropic-version: 2023-06-01" \

1083 -H "anthropic-version: 2026-10-21" \

10841084 -H "content-type: application/json" \

10851085 -d '{

10861086 "requests": [

from line 1590

15901590 ```bash cURL

15911591 curl https://api.anthropic.com/v1/messages/batches \

15921592 -H "x-api-key: $ANTHROPIC_API_KEY" \

1593 -H "anthropic-version: 2023-06-01" \

1593 -H "anthropic-version: 2026-10-21" \

15941594 -H "anthropic-beta: output-300k-2026-03-24" \

15951595 -H "content-type: application/json" \

15961596 -d '{

build-with-claude/cache-diagnostics Changed · +5 / -5 lines

from line 43

4343

4444## Basic usage

4545

46Include the `diagnostics` object on every turn. The object is the opt-in: the API stores a fingerprint only for requests that include it. On the first turn, pass `"previous_message_id": null` to opt in without a prior message to compare against. On subsequent turns, pass the `id` from the previous response. The `cache-diagnosis-2026-04-07` beta header is no longer required, and requests that still send it work as before.

46Include the `diagnostics` object on every turn. The object is the opt-in: the API stores a fingerprint only for requests that include it. On the first turn, pass `"previous_message_id": null` to opt in without a prior message to compare against. On subsequent turns, pass the `id` from the previous response. The `cache-diagnosis-2026-04-07` beta header is no longer required, and requests that send it work as before.

4747

4848<CodeGroup>

4949 ```bash cURL

from line 50

5050 # Turn 1: establish the cache and opt in to diagnostics

5151 curl -sS --fail-with-body https://api.anthropic.com/v1/messages \

5252 -H "x-api-key: $ANTHROPIC_API_KEY" \

53 -H "anthropic-version: 2023-06-01" \

53 -H "anthropic-version: 2026-10-21" \

5454 -H "content-type: application/json" \

5555 -d '{

5656 "model": "claude-opus-5-5",

from line 65

6565 # In the response, diagnostics: null means no divergence was found.

6666 curl -sS --fail-with-body https://api.anthropic.com/v1/messages \

6767 -H "x-api-key: $ANTHROPIC_API_KEY" \

68 -H "anthropic-version: 2023-06-01" \

68 -H "anthropic-version: 2026-10-21" \

6969 -H "content-type: application/json" \

7070 -d '{

7171 "model": "claude-opus-5-5",

from line 414

414414 # a null value means no divergence was found.

415415 curl -sS --fail-with-body https://api.anthropic.com/v1/messages \

416416 -H "x-api-key: $ANTHROPIC_API_KEY" \

417 -H "anthropic-version: 2023-06-01" \

417 -H "anthropic-version: 2026-10-21" \

418418 -H "content-type: application/json" \

419419 -d '{

420420 "model": "claude-opus-5-5",

from line 1035

10351035

10361036## Limitations

10371037

1038* **Claude API only:** Not available on Amazon Bedrock or Claude Models on Google Cloud (Google managed).

1038* **Claude API only:** Not available on Claude Models on Google Cloud (Google managed) or Amazon Bedrock.

10391039* **Limited retention:** Fingerprints for `previous_message_id` lookup expire after a short period. Run diagnostic comparisons between closely spaced requests.

10401040* **Same workspace:** The previous request must have run in the same organization and workspace. To check, compare the `anthropic-workspace-id` [response header](https://platform.claude.com/docs/en/api/overview#response-headers) on the two responses.

10411041* **Comparison horizon:** For very long conversations where the only change is deep in the message list, the response might be `unavailable` rather than a precise location.

build-with-claude/citations Changed · +8 / -8 lines

from line 49

4949 curl https://api.anthropic.com/v1/messages \

5050 -H "content-type: application/json" \

5151 -H "x-api-key: $ANTHROPIC_API_KEY" \

52 -H "anthropic-version: 2023-06-01" \

52 -H "anthropic-version: 2026-10-21" \

5353 -d '{

5454 "model": "claude-opus-5-5",

5555 "max_tokens": 1024,

from line 480

480480 curl https://api.anthropic.com/v1/messages \

481481 -H "content-type: application/json" \

482482 -H "x-api-key: $ANTHROPIC_API_KEY" \

483 -H "anthropic-version: 2023-06-01" \

483 -H "anthropic-version: 2026-10-21" \

484484 -d '{

485485 "model": "claude-opus-5-5",

486486 "max_tokens": 1024,

from line 832

832832 ```bash cURL

833833 curl -X POST https://api.anthropic.com/v1/messages \

834834 -H "x-api-key: $ANTHROPIC_API_KEY" \

835 -H "anthropic-version: 2023-06-01" \

835 -H "anthropic-version: 2026-10-21" \

836836 -H "content-type: application/json" \

837837 -d '{

838838 "model": "claude-opus-5-5",

from line 1085

10851085 curl https://api.anthropic.com/v1/messages \

10861086 -H "content-type: application/json" \

10871087 -H "x-api-key: $ANTHROPIC_API_KEY" \

1088 -H "anthropic-version: 2023-06-01" \

1088 -H "anthropic-version: 2026-10-21" \

10891089 -d '{

10901090 "model": "claude-opus-5-5",

10911091 "max_tokens": 1024,

from line 1384

13841384 curl https://api.anthropic.com/v1/messages \

13851385 -H "content-type: application/json" \

13861386 -H "x-api-key: $ANTHROPIC_API_KEY" \

1387 -H "anthropic-version: 2023-06-01" \

1387 -H "anthropic-version: 2026-10-21" \

13881388 -d '{

13891389 "model": "claude-opus-5-5",

13901390 "max_tokens": 1024,

from line 1660

16601660 ```bash cURL

16611661 curl -X POST https://api.anthropic.com/v1/messages \

16621662 -H "x-api-key: $ANTHROPIC_API_KEY" \

1663 -H "anthropic-version: 2023-06-01" \

1663 -H "anthropic-version: 2026-10-21" \

16641664 -H "content-type: application/json" \

16651665 -d '{

16661666 "model": "claude-opus-5-5",

from line 1909

19091909 curl https://api.anthropic.com/v1/messages \

19101910 -H "content-type: application/json" \

19111911 -H "x-api-key: $ANTHROPIC_API_KEY" \

1912 -H "anthropic-version: 2023-06-01" \

1912 -H "anthropic-version: 2026-10-21" \

19131913 -d '{

19141914 "model": "claude-opus-5-5",

19151915 "max_tokens": 1024,

from line 2346

23462346

23472347<CardGroup cols={2}>

23482348 <Card title="Streaming messages" icon="wifi-high" href="https://platform.claude.com/docs/en/build-with-claude/streaming">

2349 Handle the `citations_delta` delta type alongside text deltas to render cited responses as they stream.

2349 Learn the event flow and the other content block delta types that arrive alongside `citations_delta`.

23502350 </Card>

23512351

23522352 <Card title="Search results" icon="book-bookmark" href="https://platform.claude.com/docs/en/build-with-claude/search-results">

build-with-claude/claude-in-amazon-bedrock Changed · +19 / -26 lines

from line 12

1212

1313## Access

1414

15Amazon Bedrock sets access criteria for each Claude model individually. Claude Fable 5.1, Claude Fable 5, Claude Opus 4.8, Claude Sonnet 5, Claude Opus 4.7, and Claude Haiku 4.5 are open to all Amazon Bedrock customers. For any other model’s current criteria, check [Amazon Bedrock model access](https://console.aws.amazon.com/bedrock/home#/modelaccess) in the AWS console. Claude Mythos Preview requires an invitation through [Project Glasswing](https://anthropic.com/glasswing). For region availability, see [Regions](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#regions).

15Amazon Bedrock sets access criteria for each Claude model individually. Claude Fable 5.1, Claude Fable 5, Claude Opus 4.8, Claude Sonnet 5, Claude Opus 4.7, and Claude Haiku 4.5 are open to all Amazon Bedrock customers. For any other model’s criteria, check [Amazon Bedrock model access](https://console.aws.amazon.com/bedrock/home#/modelaccess) in the AWS console. For region availability, see [Regions](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#regions).

1616

1717## Prerequisites

1818

from line 21

2121* An AWS account with [Amazon Bedrock model access](https://console.aws.amazon.com/bedrock/home#/modelaccess) enabled for the Claude models you intend to use

2222* The [AWS CLI](https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html) installed and configured (optional, for credential management)

2323

24Claude Mythos Preview additionally requires a dedicated AWS account that has been allowlisted by the Bedrock Marketplace team. Your Anthropic account executive can submit your account ID for allowlisting (typically processed within 24 hours), and AWS sends a welcome email once it’s complete.

25

2624## Authentication

2725

2826Claude in Amazon Bedrock supports three authentication paths. Choose the one that best fits your security requirements.

from line 151

153151 -u "$AWS_ACCESS_KEY_ID:$AWS_SECRET_ACCESS_KEY" \

154152 -H "x-amz-security-token: $AWS_SESSION_TOKEN" \

155153 -H "content-type: application/json" \

156 -H "anthropic-version: 2023-06-01" \

154 -H "anthropic-version: 2026-10-21" \

157155 -d '{

158156 "model": "anthropic.claude-opus-5-5",

159157 "max_tokens": 1024,

from line 330

332330

333331Model IDs in Claude in Amazon Bedrock carry an `anthropic.` provider prefix. Model capabilities and behaviors are documented on the [Models overview](https://platform.claude.com/docs/en/models/overview) page.

334332

335| Model | Model ID | Access |

336| :------------------------------------------------------------------------------------------ | :-------------------------------- | :-------------------------------------------------------------------------------------------------- |

337| Claude Fable 5.1 | `anthropic.claude-fable-5-1` | Open |

338| Claude Mythos 5.1 ([limited availability](https://support.claude.com/en/articles/14604842)) | `anthropic.claude-mythos-5-1` | Active (verification required) |

339| Claude Fable 5 | `anthropic.claude-fable-5` | Open |

340| Claude Mythos 5 ([limited availability](https://support.claude.com/en/articles/14604842)) | `anthropic.claude-mythos-5` | Active (verification required) |

341| Claude Mythos Preview ([limited availability](https://anthropic.com/glasswing)) | `anthropic.claude-mythos-preview` | Invitation only |

342| Claude Opus 5.5 | `anthropic.claude-opus-5-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

343| Claude Opus 5 | `anthropic.claude-opus-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

344| Claude Opus 4.8 | `anthropic.claude-opus-4-8` | Open |

345| Claude Opus 4.7 | `anthropic.claude-opus-4-7` | Open |

346| Claude Sonnet 5.5 | `anthropic.claude-sonnet-5-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

347| Claude Sonnet 5 | `anthropic.claude-sonnet-5` | Open |

348| Claude Haiku 5.5 | `anthropic.claude-haiku-5-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

349| Claude Haiku 4.5 | `anthropic.claude-haiku-4-5` | Open |

333| Model | Model ID | Access |

334| :------------------------------------------------------------------------------------------ | :---------------------------- | :-------------------------------------------------------------------------------------------------- |

335| Claude Fable 5.1 | `anthropic.claude-fable-5-1` | Open |

336| Claude Mythos 5.1 ([limited availability](https://support.claude.com/en/articles/14604842)) | `anthropic.claude-mythos-5-1` | Active (verification required) |

337| Claude Fable 5 | `anthropic.claude-fable-5` | Open |

338| Claude Mythos 5 ([limited availability](https://support.claude.com/en/articles/14604842)) | `anthropic.claude-mythos-5` | Active (verification required) |

339| Claude Opus 5.5 | `anthropic.claude-opus-5-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

340| Claude Opus 5 | `anthropic.claude-opus-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

341| Claude Opus 4.8 | `anthropic.claude-opus-4-8` | Open |

342| Claude Opus 4.7 | `anthropic.claude-opus-4-7` | Open |

343| Claude Sonnet 5.5 | `anthropic.claude-sonnet-5-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

344| Claude Sonnet 5 | `anthropic.claude-sonnet-5` | Open |

345| Claude Haiku 5.5 | `anthropic.claude-haiku-5-5` | [See Access](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock#access) |

346| Claude Haiku 4.5 | `anthropic.claude-haiku-4-5` | Open |

350347

351348Use the latest version of Claude Code with Claude in Amazon Bedrock; run `claude update` to upgrade.

352349

from line 382

385382* **Global:** dynamic routing across all available regions for maximum availability. No pricing premium.

386383* **Regional:** the endpoint resolves to the single AWS region you specify, for data-residency requirements. Regional endpoints carry a 10 percent pricing premium over global endpoints. To route across multiple regions within a geography, use an [inference profile](https://docs.aws.amazon.com/bedrock/latest/userguide/cross-region-inference.html) (US, EU, JP, or AU). Regions marked **In-region only** in the table support direct single-region routing without an inference profile.

387384

388The global endpoint is available for Claude Fable 5.1, Claude Fable 5, Claude Opus 5.5, Claude Opus 5, Claude Opus 4.8, Claude Opus 4.7, Claude Sonnet 5.5, Claude Sonnet 5, Claude Haiku 5.5, and Claude Haiku 4.5. For Claude Fable 5.1, regional endpoints are currently available in `us-east-1` only. Claude Mythos Preview is regional only and is available in `us-east-1`.

385The global endpoint is available for Claude Fable 5.1, Claude Fable 5, Claude Opus 5.5, Claude Opus 5, Claude Opus 4.8, Claude Opus 4.7, Claude Sonnet 5.5, Claude Sonnet 5, Claude Haiku 5.5, and Claude Haiku 4.5. For Claude Fable 5.1, regional endpoints are available in `us-east-1` only.

389386

390387| AWS region | Location | Endpoint types |

391388| ---------------- | ------------------------- | -------------------------- |

from line 416

419416

420417## Quotas

421418

422Default quota is 2 million input tokens per minute (TPM). You can request up to 5 million input TPM and 500,000 output TPM without additional Anthropic approval. AWS enforces requests-per-minute (RPM) limits on the Bedrock side; contact AWS support for RPM adjustments.

419Default quota is 2 million input tokens per minute (TPM). You can request up to 5 million input TPM and 500,000 output TPM without additional Anthropic approval. AWS enforces requests-per-minute (RPM) limits on the Bedrock side; contact AWS Support for RPM adjustments.

423420

424421## Data retention

425422

from line 428

431428

432429## Support

433430

434For support, contact **[[email protected]](mailto:[email protected])**. Include your AWS account ID and the `request-id` from any failed API responses.

435

436<Note>

437 **Claude Mythos Preview** is a research preview model available to invited customers on Amazon Bedrock. For more information, see [Project Glasswing](https://anthropic.com/glasswing).

438</Note>

431For support, contact AWS Support or your AWS account team. Include your AWS account ID and the `request-id` from any failed API responses.

439432

build-with-claude/claude-models-on-google-cloud/anthropic-managed Changed · +66 / -42 lines

from line 249

249249 curl "https://claude.googleapis.com/v1/projects/$ANTHROPIC_GOOGLE_CLOUD_PROJECT/locations/global/workspaces/$ANTHROPIC_GOOGLE_CLOUD_WORKSPACE_ID/invoke/v1/messages" \

250250 -H "Authorization: Bearer $(gcloud auth application-default print-access-token)" \

251251 -H "content-type: application/json" \

252 -H "anthropic-version: 2023-06-01" \

252 -H "anthropic-version: 2026-10-21" \

253253 -d '{

254254 "model": "claude-opus-5-5",

255255 "max_tokens": 1024,

from line 500

500500

501501This section lists features that work on Claude Models on Google Cloud (Anthropic managed) and features that aren’t supported there. It doesn’t list every Claude API feature, and it doesn’t say whether an unlisted feature works.

502502

503The following features work on Claude Models on Google Cloud (Anthropic managed) as they do on the Claude API:

503The Models API and the following features are available on Claude Models on Google Cloud (Anthropic managed):

504504

505* **Fast mode:** [Fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode) for Claude Opus 5.5.

505* [Advisor tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/advisor-tool) (Beta)

506* [Agent Skills](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview)

507 * [Custom Skills](https://platform.claude.com/docs/en/build-with-claude/skills-guide#managing-custom-skills)

508 * [Pre-built Agent Skills](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview#pre-built-agent-skills)

509* [Batch processing](https://platform.claude.com/docs/en/build-with-claude/batch-processing)

510* [Classifier refusals](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback)

511 * [Client-side fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#client-side-fallback) (Beta)

512 * [Server-side fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#server-side-fallback) (Beta)

513* [Code execution tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool)

514* [Custom tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/define-tools)

515* [Data residency](https://platform.claude.com/docs/en/manage-claude/data-residency)

516* [Effort](https://platform.claude.com/docs/en/build-with-claude/effort)

517 * [Per-message effort](https://platform.claude.com/docs/en/build-with-claude/effort#change-effort-mid-conversation-beta) (Beta)

518* [Fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode)

519* [Files API](https://platform.claude.com/docs/en/build-with-claude/files)

520* [MCP connector](https://platform.claude.com/docs/en/agents-and-tools/mcp-connector) (Beta)

521 * [Client-side MCP helpers](https://platform.claude.com/docs/en/agents-and-tools/mcp-connector#client-side-mcp-helpers) (Beta)

522* [Memory tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/memory-tool)

523 * [Memory tool helpers](https://platform.claude.com/docs/en/agents-and-tools/tool-use/memory-tool#implement-the-memory-handler) (Beta)

524* [Message streaming](https://platform.claude.com/docs/en/build-with-claude/streaming)

525* [Message threads](https://platform.claude.com/docs/en/build-with-claude/message-threads) (Beta)

526* [Mid-conversation system messages](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages)

527 * [Mid-conversation tool changes](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages#mid-conversation-tool-changes) (Beta)

528 * [Turn-scoped system messages](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages#turn-scoped-system-messages) (Beta)

529* [OpenTelemetry](https://platform.claude.com/docs/en/manage-claude/opentelemetry) (Beta)

530* [PDF support](https://platform.claude.com/docs/en/build-with-claude/pdf-support)

531* [Prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching)

532 * [1-hour duration](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#1-hour-cache-duration)

533 * [5-minute duration (default)](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#how-prompt-caching-works)

534 * [Automatic caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#automatic-caching)

535* [SDK middleware](https://platform.claude.com/docs/en/cli-sdks-libraries/middleware)

536* [Structured outputs](https://platform.claude.com/docs/en/build-with-claude/structured-outputs)

537 * [Typed structured output helpers](https://platform.claude.com/docs/en/build-with-claude/structured-outputs#usage)

538* [Thinking](https://platform.claude.com/docs/en/build-with-claude/thinking)

539 * [Adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking-steering-and-cost)

540 * [Extended thinking](https://platform.claude.com/docs/en/build-with-claude/extended-thinking)

541* [Token counting](https://platform.claude.com/docs/en/build-with-claude/token-counting)

542* [Tool runner](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-runner) (Beta)

543* [Vision (images)](https://platform.claude.com/docs/en/build-with-claude/vision)

544* [Web fetch tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool)

545* [Web search tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool)

506546

507* **Beta features:** If a beta is limited to selected organizations on the Claude API, the Anthropic organization that setup creates needs the same access.

547If a beta feature is limited to selected organizations on the Claude API, the Anthropic organization that setup creates needs the same access.

508548

509* **Tools:** [Tool use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview), including:

510

511 * Web search

512 * Web fetch

513 * Code execution

514 * The memory tool

515 * The advisor tool

516 * The MCP connector

517

518* **Skills and files:** [Agent Skills](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview) and the [Files API](https://platform.claude.com/docs/en/build-with-claude/files).

519

520* **Model capabilities:**

521

522 * [Prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching)

523 * [Extended thinking](https://platform.claude.com/docs/en/build-with-claude/thinking)

524 * [Adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking#configuring-thinking)

525 * [Effort](https://platform.claude.com/docs/en/build-with-claude/effort)

526 * [Server-side fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#server-side-fallback)

527 * [Streaming](https://platform.claude.com/docs/en/build-with-claude/streaming)

528 * [Structured outputs](https://platform.claude.com/docs/en/build-with-claude/structured-outputs)

529 * [Token counting](https://platform.claude.com/docs/en/build-with-claude/token-counting)

530 * [Mid-conversation system messages](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages)

531

532* **Batches and models:** The [Message Batches API](https://platform.claude.com/docs/en/build-with-claude/batch-processing) and the Models API.

533

534549To turn the [Compliance API](https://platform.claude.com/docs/en/manage-claude/compliance-api) on or off for the Anthropic organization that setup creates, [open the Claude Console](https://platform.claude.com/docs/en/build-with-claude/claude-models-on-google-cloud/anthropic-managed#using-the-claude-console), go to **Settings > Compliance**, and use the **Compliance API** toggle.

535550

536551Customer-managed encryption keys (CMEK) are available with Google Cloud KMS keys. To activate CMEK for the Anthropic organization that setup creates, see [Who can use CMEK](https://platform.claude.com/docs/en/manage-claude/cmek#availability). To set up a key, follow [Set up CMEK with Google Cloud KMS](https://platform.claude.com/docs/en/manage-claude/cmek-google-cloud-kms). Then register, verify, and attach the key in the [Claude Console](https://platform.claude.com/docs/en/build-with-claude/claude-models-on-google-cloud/anthropic-managed#using-the-claude-console), as the setup page’s **Claude Platform** tab describes.

from line 552

537552

538553### Claude Managed Agents

539554

540[Claude Managed Agents](https://platform.claude.com/docs/en/managed-agents/overview) is available through Claude Models on Google Cloud (Anthropic managed), including the following:

555[Claude Managed Agents](https://platform.claude.com/docs/en/managed-agents/overview) is available through Claude Models on Google Cloud (Anthropic managed), including [sessions](https://platform.claude.com/docs/en/managed-agents/sessions) and the following features:

541556

542* [Agents](https://platform.claude.com/docs/en/managed-agents/agent-setup)

543* [Environments](https://platform.claude.com/docs/en/managed-agents/environments)

544* [Sessions](https://platform.claude.com/docs/en/managed-agents/sessions)

545* [Credential vaults](https://platform.claude.com/docs/en/managed-agents/vaults)

546* [Memory stores](https://platform.claude.com/docs/en/managed-agents/memory)

547* [Scheduled deployments](https://platform.claude.com/docs/en/managed-agents/scheduled-deployments)

548* [Self-hosted sandboxes](https://platform.claude.com/docs/en/managed-agents/self-hosted-sandboxes)

557* [Agent configurations](https://platform.claude.com/docs/en/managed-agents/agent-setup) (Beta)

558* [Cloud sandboxes](https://platform.claude.com/docs/en/managed-agents/environments) (Beta)

559* [Event deltas](https://platform.claude.com/docs/en/managed-agents/event-deltas) (Beta)

560* [GitHub access](https://platform.claude.com/docs/en/managed-agents/github) (Beta)

561* [Memory stores](https://platform.claude.com/docs/en/managed-agents/memory) (Beta)

562* [Multiagent orchestration](https://platform.claude.com/docs/en/managed-agents/multiagent-orchestration) (Beta)

563 * [Dynamic workflows](https://platform.claude.com/docs/en/managed-agents/workflow-runs) (Beta)

564 * [Session threads](https://platform.claude.com/docs/en/managed-agents/session-threads) (Beta)

565* [Outcomes](https://platform.claude.com/docs/en/managed-agents/define-outcomes) (Beta)

566* [Permission policies](https://platform.claude.com/docs/en/managed-agents/permission-policies) (Beta)

567* [Preserved thinking](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking)

568* [Scheduled deployments](https://platform.claude.com/docs/en/managed-agents/scheduled-deployments) (Beta)

569* [Self-hosted sandboxes](https://platform.claude.com/docs/en/managed-agents/self-hosted-sandboxes) (Beta)

570* [Session budgets](https://platform.claude.com/docs/en/managed-agents/budgets) (Beta)

571* [Vaults](https://platform.claude.com/docs/en/managed-agents/vaults) (Beta)

572* [Web domain restrictions](https://platform.claude.com/docs/en/managed-agents/tools-web-restrictions) (Beta)

549573

550574Access is checked when a session or a deployment is created, not while it runs:

551575

from line 666

642666 -H "Authorization: Bearer $(gcloud auth print-access-token)" \

643667 -H "Content-Type: application/json" \

644668 -d '{

645 "anthropic_version": "vertex-2023-10-16",

669 "anthropic_version": "2026-10-21",

646670 "messages": [{"role": "user", "content": "Hey Claude!"}],

647671 "max_tokens": 100

648672 }'

from line 765

741765 curl "https://claude.googleapis.com/v1/projects/$ANTHROPIC_GOOGLE_CLOUD_PROJECT/locations/global/workspaces/$ANTHROPIC_GOOGLE_CLOUD_WORKSPACE_ID/invoke/v1/messages" \

742766 -H "Authorization: Bearer $(gcloud auth application-default print-access-token)" \

743767 -H "content-type: application/json" \

744 -H "anthropic-version: 2023-06-01" \

768 -H "anthropic-version: 2026-10-21" \

745769 -d '{

746770 "model": "claude-opus-5-5",

747771 "max_tokens": 1024,

build-with-claude/claude-models-on-google-cloud/google-managed Changed · +4 / -4 lines

from line 100

100100 -H "Authorization: Bearer $(gcloud auth print-access-token)" \

101101 -H "Content-Type: application/json" \

102102 -d '{

103 "anthropic_version": "vertex-2023-10-16",

103 "anthropic_version": "2026-10-21",

104104 "messages": [{"role": "user", "content": "Hey Claude!"}],

105105 "max_tokens": 100

106106 }'

from line 353

353353 -H "Authorization: Bearer $(gcloud auth print-access-token)" \

354354 -H "Content-Type: application/json" \

355355 -d '{

356 "anthropic_version": "vertex-2023-10-16",

356 "anthropic_version": "2026-10-21",

357357 "messages": [{"role": "user", "content": "Hey Claude!"}],

358358 "max_tokens": 100

359359 }'

from line 549

549549 curl "https://claude.googleapis.com/v1/projects/$ANTHROPIC_GOOGLE_CLOUD_PROJECT/locations/global/workspaces/$ANTHROPIC_GOOGLE_CLOUD_WORKSPACE_ID/invoke/v1/messages" \

550550 -H "Authorization: Bearer $(gcloud auth application-default print-access-token)" \

551551 -H "content-type: application/json" \

552 -H "anthropic-version: 2023-06-01" \

552 -H "anthropic-version: 2026-10-21" \

553553 -d '{

554554 "model": "claude-opus-5-5",

555555 "max_tokens": 1024,

from line 631

631631 -H "Authorization: Bearer $(gcloud auth print-access-token)" \

632632 -H "Content-Type: application/json" \

633633 -d '{

634 "anthropic_version": "vertex-2023-10-16",

634 "anthropic_version": "2026-10-21",

635635 "messages": [{"role": "user", "content": "Hey Claude!"}],

636636 "max_tokens": 100

637637 }'

build-with-claude/claude-on-amazon-bedrock-legacy Changed · +30 / -16 lines

from line 10

1010

1111<Warning>

1212 The latest version of the SDK sends API version `2026-10-21` by default, including through Amazon Bedrock. The ARN-versioned models in the model table on this page don’t support `2026-10-21` on Bedrock, and requests to them that send it can return a 400 error. If you call them through the SDK, keep the SDK on the version you use today, and pin that version where you install it so a later install doesn’t upgrade it. The Boto3 example sets `anthropic_version` in its request body and isn’t affected.

13

14 To install an SDK for these models, use the commands in [Install an SDK for accessing Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy#install-an-sdk-for-accessing-bedrock), which install the last major version before API version `2026-10-21`.

1315</Warning>

1416

1517Calling Claude through Bedrock slightly differs from how you would call Claude on the Claude API directly. This guide walks you through completing an API call to Claude on Bedrock using one of Anthropic’s [client SDKs](https://platform.claude.com/docs/en/cli-sdks-libraries/overview).

from line 35

3335<Tabs>

3436 <Tab title="Python">

3537 ```bash

36 pip install -U "anthropic[bedrock]"

38 pip install -U "anthropic[bedrock]<2"

3739 ```

3840 </Tab>

3941

4042 <Tab title="TypeScript">

4143 ```bash

42 npm install @anthropic-ai/bedrock-sdk

44 npm install @anthropic-ai/bedrock-sdk@0

4345 ```

46

47 The example in [List available models](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy#list-available-models) also needs AWS’s Bedrock client. Install it with `npm install @aws-sdk/client-bedrock`.

4448 </Tab>

4549

4650 <Tab title="C#">

4751 ```bash

48 dotnet add package Anthropic.Bedrock

52 dotnet add package Anthropic.Bedrock --version 0.14.1

4953 ```

54

55 If your project also references the `Anthropic` package, keep it below version 13.

56

57 The example in [List available models](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy#list-available-models) also needs AWS’s Bedrock client. Install it with `dotnet add package AWSSDK.Bedrock`.

5058 </Tab>

5159

5260 <Tab title="Go">

from line 61

5361 ```bash

5462 go get github.com/anthropics/anthropic-sdk-go/bedrock

5563 ```

64

65 The example in [List available models](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy#list-available-models) also needs AWS’s Bedrock client. Install it with `go get github.com/aws/aws-sdk-go-v2/service/bedrock`.

5666 </Tab>

5767

5868 <Tab title="Java">

5969 <CodeGroup exclude="shell, python, typescript, csharp, go, php, ruby">

6070 ```groovy Gradle

61 implementation("com.anthropic:anthropic-java:3.4.0")

62 implementation("com.anthropic:anthropic-java-bedrock:3.4.0")

71 implementation("com.anthropic:anthropic-java:2.+")

72 implementation("com.anthropic:anthropic-java-bedrock:2.+")

6373 ```

6474

6575 ```xml Maven

from line 76

6676 <dependency>

6777 <groupId>com.anthropic</groupId>

6878 <artifactId>anthropic-java</artifactId>

69 <version>3.4.0</version>

79 <version>[2.0.0,3.0.0)</version>

7080 </dependency>

7181 <dependency>

7282 <groupId>com.anthropic</groupId>

7383 <artifactId>anthropic-java-bedrock</artifactId>

74 <version>3.4.0</version>

84 <version>[2.0.0,3.0.0)</version>

7585 </dependency>

7686 ```

7787

from line 113

103113 }

104114 ```

105115 </CodeGroup>

116

117 The example in [List available models](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy#list-available-models) also needs AWS’s Bedrock client. Add `software.amazon.awssdk:bedrock` to your build.

106118 </Tab>

107119

108120 <Tab title="PHP">

109121 ```bash

110 composer require anthropic-ai/sdk aws/aws-sdk-php

122 composer require "anthropic-ai/sdk:<1.0" aws/aws-sdk-php

111123 ```

112124 </Tab>

113125

from line 126

114126 <Tab title="Ruby">

115127 ```bash

116128 # Gemfile

117 gem "anthropic"

129 gem "anthropic", "~> 1.0"

118130 gem "aws-sdk-bedrockruntime"

119131 ```

132

133 The example in [List available models](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy#list-available-models) also needs AWS’s Bedrock client. Add `gem "aws-sdk-bedrock"` to your `Gemfile`.

120134 </Tab>

121135

122136 <Tab title="Boto3 (Python)">

from line 664

650664 import com.anthropic.client.AnthropicClient;

651665 import com.anthropic.client.okhttp.AnthropicOkHttpClient;

652666 import com.anthropic.models.messages.MessageCreateParams;

667 import software.amazon.awssdk.regions.Region;

653668

654669 // Option 1: Set AWS_BEARER_TOKEN_BEDROCK environment variable and use fromEnv()

655670 AnthropicClient client = AnthropicOkHttpClient.builder()

from line 675

660675 client = AnthropicOkHttpClient.builder()

661676 .backend(BedrockBackend.builder()

662677 .apiKey("your-bearer-token")

678 .region(Region.US_WEST_2)

663679 .build())

664680 .build();

665681

from line 1031

10151031 using Anthropic.Bedrock;

10161032 using Anthropic.Models.Messages;

10171033

1034 // Default credentials resolve region from AWS_REGION env var

10181035 AnthropicBedrockClient client = new(

1019 new AnthropicBedrockPrivateKeyCredentials { Region = "us-west-2" }

1036 await AnthropicBedrockCredentialsHelper.FromEnv()

1037 ?? throw new InvalidOperationException("AWS credentials not configured.")

10201038 );

10211039

10221040 // Using US regional endpoint (CRIS)

from line 1130

11121130 ```

11131131 </Tab>

11141132</Tabs>

1115

1116<Note>

1117 **Claude Mythos Preview** is a research preview model available to invited customers on Amazon Bedrock. For more information, see [Project Glasswing](https://anthropic.com/glasswing).

1118</Note>

11191133

11201134## Additional resources

11211135

One read of Claude Developer Platformapi-20261104T020710Z

What this read moved

about-claude/models/optimizing-for-cost-and-intelligence Changed · +5 / -3 lines

about-claude/pricing Changed · +8 / -8 lines

agents-and-tools/agent-skills/quickstart Changed · +7 / -7 lines

agents-and-tools/mcp-connector Changed · +3 / -3 lines

agents-and-tools/tool-use/advisor-tool Changed · +41 / -43 lines

agents-and-tools/tool-use/bash-tool Changed · +3 / -3 lines

agents-and-tools/tool-use/browser-use-tool Changed · +11 / -11 lines

agents-and-tools/tool-use/code-execution-tool Changed · +8 / -6 lines

agents-and-tools/tool-use/computer-use-tool Changed · +5 / -5 lines

agents-and-tools/tool-use/overview Changed · +3 / -3 lines

agents-and-tools/tool-use/programmatic-tool-calling Changed · +18 / -16 lines

agents-and-tools/tool-use/server-tools Changed · +4 / -3 lines

agents-and-tools/tool-use/strict-tool-use Changed · +3 / -3 lines

agents-and-tools/tool-use/text-editor-tool Changed · +4 / -4 lines

agents-and-tools/tool-use/web-fetch-tool Changed · +4 / -4 lines

api/claude-platform-on-aws-iam-actions Changed · +29 / -24 lines

api/errors Changed · +6 / -4 lines

api/migration-2026-10-21 Changed · +7 / -7 lines

build-with-claude/batch-processing Changed · +11 / -11 lines

build-with-claude/cache-diagnostics Changed · +5 / -5 lines

build-with-claude/citations Changed · +8 / -8 lines

build-with-claude/claude-in-amazon-bedrock Changed · +19 / -26 lines

build-with-claude/claude-models-on-google-cloud/anthropic-managed Changed · +66 / -42 lines

build-with-claude/claude-models-on-google-cloud/google-managed Changed · +4 / -4 lines

build-with-claude/claude-on-amazon-bedrock-legacy Changed · +30 / -16 lines