thinking changedbuild-with-claude/thinking
Recorded here
Lines+23added
Lines−15removed
From line
29
where the diff opens
First seen
14 Aug 2026
this site's first read of the page
Recorded edits42to this page, all time
The whole hunk
from line 29, old and new numbered
/
from line 29
2929 - claude-sonnet-4-5-20250929
3030 supportedPlatforms:
3131 Claude API: stable
32 Amazon Bedrock: stable
33 Amazon Bedrock (bedrock-runtime): stable
3234 Claude Platform on AWS: stable
3335 Claude Models on Google Cloud (Anthropic managed): stable
34 Amazon Bedrock: stable
35 Amazon Bedrock (InvokeModel): stable
3636 Claude Models on Google Cloud (Google managed): stable
3737 Microsoft Foundry: stable
3838---
from line 496
496496
497497Claude Sonnet 5.5 also has thinking on by default, and it rejects `thinking: {type: "disabled"}` with a 400 error. To turn off up-front thinking, send `thinking: {type: "between_tools"}` instead. It’s the lowest thinking setting on Claude Sonnet 5.5, and it’s accepted at [effort](https://platform.claude.com/docs/en/build-with-claude/effort) `high` or below. The model still returns its [progress updates between tool calls](https://platform.claude.com/docs/en/build-with-claude/thinking#progress-updates). Without tools, the response contains only text, as with `disabled` on Claude Sonnet 5. See [Running without up-front thinking](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-sonnet-5-5#running-without-up-front-thinking) for prompting guidance.
498498
499Claude Haiku 5.5 also has thinking on by default and accepts `thinking: {type: "disabled"}` at [effort](https://platform.claude.com/docs/en/build-with-claude/effort) `high` or below. At `xhigh` or `max` effort, that combination returns a 400 error. To get less thinking, lower the effort level first. With thinking off, the model can skip a tool call it needs when you also request JSON output. A [per-message `output_config.effort`](https://platform.claude.com/docs/en/build-with-claude/effort#change-effort-mid-conversation-beta) that differs from the level in effect also returns a 400 error while thinking is off. See [Use effort to control thinking](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-haiku-5-5#use-effort-to-control-thinking) and [Use adaptive thinking with JSON output and your own tools](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-haiku-5-5#json-output-with-your-own-tools).
499Claude Haiku 5.5 also has thinking on by default and accepts `thinking: {type: "disabled"}` at [effort](https://platform.claude.com/docs/en/build-with-claude/effort) `high` or below. At `xhigh` or `max` effort, that combination returns a 400 error. To get less thinking, lower the effort level first. With thinking off, the model can skip a tool call it needs when you also request JSON output. A [per-message `output_config.effort`](https://platform.claude.com/docs/en/build-with-claude/effort-per-message) that differs from the level in effect also returns a 400 error while thinking is off. See [Use effort to control thinking](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-haiku-5-5#use-effort-to-control-thinking) and [Use adaptive thinking with JSON output and your own tools](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-haiku-5-5#json-output-with-your-own-tools).
500500
501501Claude Fable 5.2, Claude Mythos 5.2, Claude Fable 5.1, Claude Mythos 5.1, Claude Fable 5, Claude Mythos 5, Claude Opus 5.5, and Claude Mythos Preview reject `thinking: {type: "disabled"}`. Thinking can’t be turned off on these models.
502502
from line 894
894894data: {"type":"content_block_start","index":1,"content_block":{"type":"text","text":""}}
895895```
896896
897With `display: "updates"` (beta), reasoning blocks stream as they do under `"omitted"`. Each [progress-update block](https://platform.claude.com/docs/en/build-with-claude/thinking#progress-updates) streams its text as `thinking_delta` events before the `tool_use` block it introduces. A pause of several seconds before the progress-update block opens is normal:
897With `display: "updates"` (in beta), reasoning blocks stream as they do under `"omitted"`. Each [progress-update block](https://platform.claude.com/docs/en/build-with-claude/thinking#progress-updates) streams its text as `thinking_delta` events before the `tool_use` block it introduces. A pause of several seconds before the progress-update block opens is normal:
898898
899899```sse Output
900900event: content_block_start
from line 940
940940
941941Thinking works alongside [tool use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview), letting Claude reason through tool selection and process tool results. Two constraints apply:
942942
9431. **Tool choice limitation (manual mode):** tool use with manual extended thinking (`thinking: {type: "enabled"}`) only supports `tool_choice: {"type": "auto"}` (the default) or `tool_choice: {"type": "none"}`. Using `tool_choice: {"type": "any"}` or `tool_choice: {"type": "tool", "name": "..."}` results in an error because these options force tool use, which is incompatible with manual extended thinking. Adaptive thinking, including on models where thinking is on by default, supports forced tool use, except on Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.2, Claude Mythos 5.2, Claude Fable 5.1, and Claude Mythos 5.1 (see [Response prefill and forced tool use](https://platform.claude.com/docs/en/build-with-claude/thinking#limits-and-feature-compatibility)). Where forced tool use is accepted, the response starts with the tool call and has no `thinking` block.
9431. **Tool choice limitation (manual mode):** tool use with manual extended thinking (`thinking: {type: "enabled"}`) only supports `tool_choice: {"type": "auto"}` (the default) or `tool_choice: {"type": "none"}`. Using `tool_choice: {"type": "any"}` or `tool_choice: {"type": "tool", "name": "..."}` results in an error because these options force tool use, which is incompatible with manual extended thinking. Adaptive thinking, including on models where thinking is on by default, supports forced tool use, except on Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.2, Claude Mythos 5.2, Claude Fable 5.1, and Claude Mythos 5.1 (see [Response prefill and forced tool use](https://platform.claude.com/docs/en/build-with-claude/thinking#response-prefill-and-forced-tool-use)). Where forced tool use is accepted, the response starts with the tool call and has no `thinking` block.
9449442. **Preserving thinking blocks:** when you return tool results, you must pass the thinking blocks from the assistant message back to the API, complete and unmodified. See [Preserving thinking blocks](https://platform.claude.com/docs/en/build-with-claude/thinking#preserving-thinking-blocks).
945945
946946**A tool-use loop is one assistant turn.** From the model’s perspective, an assistant turn ends only when Claude completes its full response, which might include multiple tool calls and results. This whole sequence is a single assistant turn:
from line 1004
10041004 Consecutive tool calls do not require interleaved thinking. Claude can chain tool calls with or without interleaved thinking. Interleaving changes where thinking blocks appear between tool calls, not whether tool calls can chain.
10051005</Note>
10061006
1007With adaptive thinking, interleaved thinking is automatic on every model that supports adaptive thinking. No beta header is needed. On Claude Fable 5.2, Claude Mythos 5.2, Claude Fable 5.1, Claude Mythos 5.1, Claude Fable 5, Claude Mythos 5, Claude Mythos Preview, Claude Opus 5.5, Claude Opus 5, Claude Sonnet 5.5, Claude Opus 4.8, and Claude Opus 4.7, reasoning between tool calls always appears in thinking blocks. Claude Haiku 4.5 does not support interleaved thinking. On models using manual extended thinking, interleaving requires a beta header and changes how the thinking budget is counted. [Interleaved thinking in manual mode](https://platform.claude.com/docs/en/build-with-claude/extended-thinking#interleaved-thinking) covers the per-model rules and platform-specific header behavior.
1007With adaptive thinking, interleaved thinking is automatic on every model that supports adaptive thinking. On Claude Fable 5.2, Claude Mythos 5.2, Claude Fable 5.1, Claude Mythos 5.1, Claude Fable 5, Claude Mythos 5, Claude Mythos Preview, Claude Opus 5.5, Claude Opus 5, Claude Sonnet 5.5, Claude Opus 4.8, and Claude Opus 4.7, reasoning between tool calls always appears in thinking blocks. Claude Haiku 4.5 does not support interleaved thinking. On models using manual extended thinking, interleaving requires a beta header and changes how the thinking budget is counted. [Interleaved thinking in manual mode](https://platform.claude.com/docs/en/build-with-claude/extended-thinking#interleaved-thinking) covers the per-model rules and platform-specific header behavior.
10081008
10091009With interleaved thinking, the thinking allocation can span the entire assistant turn rather than a single response. Interleaved thinking is only supported for [tools used through the Messages API](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview).
10101010
from line 1022
10221022| `"updates"` (beta) | Empty `thinking` field | Summary text |
10231023| `"summarized"` | Summary text | Summary text, not distinguishable from a reasoning block |
10241024
1025On Claude Sonnet 5.5, the lowest thinking setting is `thinking: {type: "between_tools"}`. It turns off up-front thinking, and each progress update comes back with its summary text, as it would under `display: "updates"`. `between_tools` is accepted only at [effort](https://platform.claude.com/docs/en/build-with-claude/effort) `high` or below. At `xhigh` or `max`, a request with it returns a 400 error. `between_tools` takes no other field: `display`, `budget_tokens`, or `block_binding` sent with it returns a 400 error. The setting needs no beta header and works on every platform that offers Claude Sonnet 5.5. Pass the blocks back unchanged: a progress-update block you send back gives the model the full note it wrote, not the summary.
1025On Claude Sonnet 5.5, the lowest thinking setting is `thinking: {type: "between_tools"}`. It turns off up-front thinking, and each progress update comes back with its summary text, as it would under `display: "updates"`. `between_tools` is accepted only at [effort](https://platform.claude.com/docs/en/build-with-claude/effort) `high` or below. At `xhigh` or `max`, a request with it returns a 400 error. `between_tools` takes no other field: `display`, `budget_tokens`, or `block_binding` sent with it returns a 400 error. The setting works on every platform that offers Claude Sonnet 5.5. Pass the blocks back unchanged: a progress-update block you send back gives the model the full note it wrote, not the summary.
10261026
10271027Use `display: "updates"` for an agent interface that keeps reasoning hidden and shows the user a status line at each step. Under it, any `thinking` block with non-empty text is a progress update, so render those and nothing else. It’s in beta and requires the beta header `thinking-display-updates-2026-08-18` (to send it on Amazon Bedrock, Claude Models on Google Cloud (Google managed), or Microsoft Foundry, see [Beta features on other platforms](https://platform.claude.com/docs/en/api/beta-headers#beta-features-on-other-platforms)). Without it, the value is rejected with the same 400 `invalid_request_error` as an unknown `display` value.
10281028
from line 1085
10851085* The text you receive is a summary of the progress update, normally a sentence or two. Don’t rely on its length. The progress update counts toward `usage.output_tokens` at its full length, not the summary’s.
10861086* A progress-update block can come back with an empty `thinking` field under any `display` value. Render nothing for an empty block. Under `"updates"` it looks the same as an empty reasoning block and needs no separate handling.
10871087* When a response stops on `max_tokens`, `model_context_window_exceeded`, or `stop_sequence` soon after a tool call or tool result, its last block can be a progress-update block standing in for the work the model hadn’t completed. Under `"updates"` and `"summarized"` its text is exactly `This part of the response was interrupted before it finished.` and you can show it like any other update. Under `"omitted"` it’s empty. To continue, pass the assistant turn back unchanged and append a new `user` message (with a `tool_result` for each `tool_use` block in that turn).
1088* When [streaming](https://platform.claude.com/docs/en/build-with-claude/thinking#streaming-thinking), expect a pause of several seconds before a progress-update block opens. See the `"updates"` trace in [Streaming thinking](https://platform.claude.com/docs/en/build-with-claude/thinking#streaming-thinking).
1088* When streaming, expect a pause of several seconds before a progress-update block opens. See the `"updates"` trace in [Streaming thinking](https://platform.claude.com/docs/en/build-with-claude/thinking#streaming-thinking).
10891089* These models write fewer progress updates at higher [effort](https://platform.claude.com/docs/en/build-with-claude/effort) and in long tool chains. If your interface depends on them, see [Ask for user-facing progress updates](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-1#ask-for-user-facing-progress-updates). For Claude Opus 5.5 and Claude Sonnet 5.5, see User-facing progress updates in [Prompting Claude Opus 5.5](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5-5#user-facing-progress-updates) or [Prompting Claude Sonnet 5.5](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-sonnet-5-5#user-facing-progress-updates). For Claude Fable 5.2, see [Keep users informed during long turns](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-2#keep-users-informed-during-long-turns).
10901090
10911091### Thinking block preservation by model
from line 1100
11001100* **Cache optimization:** preserved thinking blocks allow cache hits during tool use, as they are passed back with tool results and cached incrementally across the assistant turn, resulting in token savings in multistep workflows.
11011101* **No intelligence impact:** preserving thinking blocks has no negative effect on model performance.
11021102
1103The tradeoff is context usage: long conversations consume more context space on keep-all models, because retained thinking blocks count as input like any other conversation history (see [Thinking and the context window](https://platform.claude.com/docs/en/build-with-claude/thinking#thinking-and-the-context-window)). The behavior is automatic in both regimes. No code changes or beta headers are required, and you should keep passing complete, unmodified thinking blocks back as described in [Preserving thinking blocks](https://platform.claude.com/docs/en/build-with-claude/thinking#preserving-thinking-blocks). To override the default in either direction, use [thinking block clearing](https://platform.claude.com/docs/en/build-with-claude/context-editing#thinking-block-clearing).
1103The tradeoff is context usage: long conversations consume more context space on keep-all models, because retained thinking blocks count as input like any other conversation history (see [Thinking and the context window](https://platform.claude.com/docs/en/build-with-claude/thinking#thinking-and-the-context-window)). The behavior is automatic in both regimes. Keep passing complete, unmodified thinking blocks back, as described in [Preserving thinking blocks](https://platform.claude.com/docs/en/build-with-claude/thinking#preserving-thinking-blocks). To override the default in either direction, use [thinking block clearing](https://platform.claude.com/docs/en/build-with-claude/context-editing#thinking-block-clearing).
11041104
1105**Switching models mid-conversation.** Keep passing thinking blocks back unchanged when you switch models, for example after a [classifier refusal fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback). A thinking block is readable only by the model that produced it and certain other models, and the API ignores or drops the blocks the target model can’t read. On Claude Fable 5.1 and Claude Mythos 5.1 the direction matters: they read earlier models’ thinking blocks, and only they and, on the Claude API, Claude Fable 5.2 and Claude Mythos 5.2 read those two models’ blocks, so switching to them keeps the conversation’s reasoning and switching from them to any model outside that group drops it (see [how dropped blocks are billed and reported](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#switching-models)). Claude Opus 5.5 reads Claude Opus 5’s thinking blocks and those of earlier Opus, Sonnet, and Haiku models, and, on the Claude API and Claude Models on Google Cloud (Google managed), of Claude Sonnet 5.5 and Claude Haiku 5.5, but not those of the Claude Fable and Claude Mythos models; on the Claude API, Claude Fable 5.1 and Claude Mythos 5.1 read Claude Opus 5.5’s blocks. A switch from Claude Opus 5.5 up to Claude Fable 5.1 on the Claude API keeps the earlier turns’ reasoning; a switch from Claude Fable 5.1 to Claude Opus 5.5 drops it. On the Claude API, Claude Fable 5.2 and Claude Mythos 5.2 read each other’s thinking blocks and those from other models, including Claude Opus 5.5, Claude Sonnet 5.5, and Claude Haiku 5.5. No other model reads their blocks, so a switch from either of them to any other model drops the earlier turns’ reasoning. On Claude Models on Google Cloud (Google managed), Claude Fable 5.2 and Claude Mythos 5.2 read the same blocks as on the Claude API. Claude Sonnet 5.5 reads thinking blocks from Claude Sonnet 5, Claude Opus 4.8, Claude Haiku 4.5, and earlier models, and, on the Claude API and Claude Models on Google Cloud (Google managed), from Claude Haiku 5.5, but not from Claude Opus 5, Claude Opus 5.5, or any Claude Fable or Claude Mythos model. On those two platforms, Claude Opus 5.5 reads Claude Sonnet 5.5’s blocks, as do Claude Fable 5.2 and Claude Mythos 5.2 on the Claude API, and no other model does: a switch from Claude Sonnet 5.5 up to Claude Opus 5.5 there, or to Claude Fable 5.2 or Claude Mythos 5.2 where they read Claude Sonnet 5.5 blocks, keeps the earlier turns’ reasoning, and any other switch away from Claude Sonnet 5.5 drops it. Claude Haiku 5.5 reads thinking blocks from Claude Sonnet 5, Claude Opus 4.8, Claude Haiku 4.5, and earlier models, but not from Claude Opus 5, Claude Opus 5.5, Claude Sonnet 5.5, or any Claude Fable or Claude Mythos model. On the Claude API and Claude Models on Google Cloud (Google managed), Claude Opus 5.5 and Claude Sonnet 5.5 read Claude Haiku 5.5’s blocks, as do Claude Fable 5.2 and Claude Mythos 5.2 on the Claude API, and no other model does. Strip prior `thinking` and `redacted_thinking` blocks yourself only to save input tokens on models that ignore rather than drop them, and never when redeeming a [fallback credit](https://platform.claude.com/docs/en/build-with-claude/fallback-credit), which requires the body unchanged.
1105**Switching models mid-conversation.** Keep passing thinking blocks back unchanged when you switch models, for example after a [classifier refusal fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback). A thinking block is readable only by the model that produced it and certain other models. The API ignores or drops the blocks that the target model can’t read (see [how dropped blocks are billed and reported](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#switching-models)). The following list says, for each model in it, what the model reads and which models read its blocks:
11061106
1107* **Claude Fable 5.2 and Claude Mythos 5.2:** On the Claude API and Claude Models on Google Cloud (Google managed), they read each other’s thinking blocks. There, they also read blocks from other models, including Claude Opus 5.5, Claude Sonnet 5.5, and Claude Haiku 5.5. No other model reads Claude Fable 5.2 or Claude Mythos 5.2 blocks. So a switch from either to a model outside this pair drops the earlier turns’ reasoning.
1108* **Claude Fable 5.1 and Claude Mythos 5.1:** They read each other’s thinking blocks and those of earlier models, except Claude Mythos Preview’s. On the Claude API and Claude Models on Google Cloud (Google managed), Claude Fable 5.2 and Claude Mythos 5.2 also read their blocks. No other model does. A switch to Claude Fable 5.1 or Claude Mythos 5.1 from a model whose blocks they read keeps the conversation’s reasoning. A switch from either to a model that can’t read their blocks drops it.
1109* **Claude Opus 5.5:** It reads thinking blocks from Claude Opus 5 and from earlier Opus, Sonnet, and Haiku models. On the Claude API and Claude Models on Google Cloud (Google managed), it also reads them from Claude Sonnet 5.5 and Claude Haiku 5.5. It doesn’t read them from the Claude Fable and Claude Mythos models. On the Claude API, Claude Fable 5.1 and Claude Mythos 5.1 read Claude Opus 5.5’s blocks. On the Claude API and Claude Models on Google Cloud (Google managed), Claude Fable 5.2 and Claude Mythos 5.2 also read them. A switch from Claude Opus 5.5 up to Claude Fable 5.1 on the Claude API keeps the earlier turns’ reasoning. A switch from Claude Fable 5.1 to Claude Opus 5.5 drops it.
1110* **Claude Sonnet 5.5:** It reads thinking blocks from Claude Sonnet 5, Claude Opus 4.8, Claude Haiku 4.5, and earlier models. On the Claude API and Claude Models on Google Cloud (Google managed), it also reads them from Claude Haiku 5.5. It doesn’t read them from Claude Opus 5, Claude Opus 5.5, or any Claude Fable or Claude Mythos model. On those two platforms, Claude Opus 5.5, Claude Fable 5.2, and Claude Mythos 5.2 read Claude Sonnet 5.5’s blocks. No other model does. A switch from Claude Sonnet 5.5 up to one of those three models there keeps the earlier turns’ reasoning. Any other switch away from Claude Sonnet 5.5 drops it.
1111* **Claude Haiku 5.5:** It reads thinking blocks from Claude Sonnet 5, Claude Opus 4.8, Claude Haiku 4.5, and earlier models. It doesn’t read them from Claude Opus 5, Claude Opus 5.5, Claude Sonnet 5.5, or any Claude Fable or Claude Mythos model. Four models read Claude Haiku 5.5’s blocks: Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.2, and Claude Mythos 5.2. They do so on the Claude API and Claude Models on Google Cloud (Google managed). No other model does.
1112
1113Strip prior `thinking` and `redacted_thinking` blocks yourself only to save input tokens on models that ignore rather than drop them. Never strip them when redeeming a [fallback credit](https://platform.claude.com/docs/en/build-with-claude/fallback-credit), which requires the body unchanged.
1114
11071115## Preserved thinking
11081116
11091117[Preserved thinking](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking) determines whether the model can use a thinking block that you send back from an earlier turn. The API checks the `signature` of every `thinking` or `redacted_thinking` block in a request for two things:
11101118
11111119* **The model that produced it.** Each model reads its own thinking blocks and those of a fixed set of other models. Claude Fable 5.1 reads blocks from Claude Opus 5 and, on the Claude API, from Claude Opus 5.5; neither Claude Opus 5 nor Claude Opus 5.5 reads blocks from Claude Fable 5.1. The API drops a block the current model can’t read, without an error and without billing it. See [Switching models mid-conversation](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#switching-models).
1112* **Everything sent before it.** A block stays valid only while the top-level `system` prompt, the `tools`, and the messages before it are unchanged. If any of them changes, that block and every later thinking block fail the prefix check. See [Keeping the prefix unchanged](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#prefix-check).
1120* **Everything sent before it.** A block stays valid only while the top-level `system` prompt, the `tools`, and the messages before it are unchanged. If any of them changes, that block and every later thinking block already in the history fail the prefix check. See [Keeping the prefix unchanged](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#prefix-check).
11131121
11141122Thinking blocks from Claude Sonnet 5.5, Claude Haiku 5.5, Claude Fable 5.2, and Claude Mythos 5.2 are also tied to the account that produced them. See [Thinking blocks stay with the account that produced them](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#account-bound-thinking) for where this applies.
11151123
from line 1134
11261134
11271135[Prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) interacts with thinking in a few specific ways. The following rules apply in both thinking modes.
11281136
1129**Configuration changes invalidate caching.** The thinking configuration and the resolved [`effort`](https://platform.claude.com/docs/en/build-with-claude/effort) level are rendered into the prompt itself, so changing any of them starts a new cache prefix. Switching between `adaptive`, `enabled`, and `disabled`, changing `budget_tokens`, and changing the effort value all invalidate cache breakpoints: message-level breakpoints always miss, and tool and system-prompt breakpoints can miss too, depending on where the model renders the configuration. Treat any thinking or top-level effort change as starting the cache over. On models that support [per-message effort](https://platform.claude.com/docs/en/build-with-claude/effort#change-effort-mid-conversation-beta), an effort change carried in a `role: "system"` message inside `messages` leaves the cached prefix intact. Consecutive requests that keep the same configuration preserve the cache, and setting a parameter explicitly to its default value is equivalent to omitting it. A thinking block the API drops under any [preserved-thinking condition](https://platform.claude.com/docs/en/build-with-claude/thinking#preserved-thinking) changes the cached prefix from that block’s position onward. Blocks passed back unchanged keep the cache intact. A worked demonstration with usage output is on the [Steering thinking](https://platform.claude.com/docs/en/build-with-claude/thinking-steering-and-cost#prompt-caching) page.
1137**Configuration changes invalidate caching.** The thinking configuration and the resolved [`effort`](https://platform.claude.com/docs/en/build-with-claude/effort) level are rendered into the prompt itself, so changing any of them starts a new cache prefix. Switching between `adaptive`, `enabled`, and `disabled`, changing `budget_tokens`, and changing the effort value all invalidate cache breakpoints: message-level breakpoints always miss, and tool and system-prompt breakpoints can miss too, depending on where the model renders the configuration. Treat any thinking or top-level effort change as starting the cache over. On models that support [per-message effort](https://platform.claude.com/docs/en/build-with-claude/effort-per-message), an effort change carried in a `role: "system"` message inside `messages` leaves the cached prefix intact. Consecutive requests that keep the same configuration preserve the cache, and setting a parameter explicitly to its default value is equivalent to omitting it. A thinking block the API drops under any [preserved-thinking condition](https://platform.claude.com/docs/en/build-with-claude/thinking#preserved-thinking) changes the cached prefix from that block’s position onward. Blocks passed back unchanged keep the cache intact. A worked demonstration with usage output is on the [Steering thinking](https://platform.claude.com/docs/en/build-with-claude/thinking-steering-and-cost#prompt-caching) page.
11301138
11311139**Thinking blocks are cached with tool results.** During a tool-use loop, caching occurs when you make a follow-up request that includes tool results. At that point the previous conversation history, including its thinking blocks, can be cached, and those cached thinking blocks count as input tokens in your usage metrics when read from the cache. This occurs automatically, even without explicit `cache_control` markers, and behaves the same for regular and interleaved thinking. The tradeoff: thinking blocks you never see again in responses still contribute to input token usage when read from cache.
11321140
from line 1168
11601168
11611169## Thinking and the context window
11621170
1163`max_tokens`, which includes all thinking Claude generates in the current turn, is enforced as a strict limit. On Claude 4.5 models and newer, if input tokens plus `max_tokens` exceeds the context window size, the API accepts the request. If generation then reaches the context window limit, it stops with `stop_reason: "model_context_window_exceeded"` instead of returning an error. On earlier models, the API returns a validation error instead. See [Handling stop reasons](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons).
1171`max_tokens`, which includes all thinking Claude generates in the current turn, is enforced as a strict limit. On Claude 4.5 models and newer, if input tokens plus `max_tokens` exceeds the context window size, the API accepts the request. If generation then reaches the context window limit, it stops with `stop_reason: "model_context_window_exceeded"` instead of returning an error. On earlier models, the API returns a validation error instead. See [Stop reasons and fallback](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons).
11641172
11651173How thinking counts against the window depends on when it was generated:
11661174
from line 1205
11971205* When [streaming responses](https://platform.claude.com/docs/en/build-with-claude/thinking#streaming-thinking), the signature arrives as a `signature_delta` inside a `content_block_delta` event just before the `content_block_stop` event.
11981206* `signature` values are significantly longer in Claude 4 and later models than in previous models.
11991207* The `signature` field is opaque: don’t interpret or parse it.
1200* `signature` values are compatible across platforms, including the Claude API, [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). Values generated on one platform work on another, except that thinking blocks from Claude Sonnet 5.5, Claude Haiku 5.5, Claude Fable 5.2, and Claude Mythos 5.2 work only in the account that produced them or in an account linked to it. See [Thinking blocks stay with the account that produced them](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#account-bound-thinking).
1208* `signature` values are compatible across platforms, including the Claude API, [Amazon Bedrock (`bedrock-mantle`)](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). Values generated on one platform work on another, except that thinking blocks from Claude Sonnet 5.5, Claude Haiku 5.5, Claude Fable 5.2, and Claude Mythos 5.2 work only in the account that produced them or in an account linked to it. See [Thinking blocks stay with the account that produced them](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#account-bound-thinking).
12011209
12021210## Redacted thinking blocks
12031211
No line in this hunk matches that.