Batches changedapi/messages/batches
Nearest release: v2.1.283, published 3 hours after this site recorded the change. Shown because the two are within 24 hours of each other. Nothing here says the release caused the edit.
Recorded here
Lines+326added
Lines−333removed
From line
33
where the diff opens
First seen
14 Aug 2026
this site's first read of the page
Recorded edits13to this page, all time
The whole hunk
from line 33, old and new numbered
/
This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.
from line 33
3333
3434 List of requests for prompt completion. Each is an individual request to create a Message.
3535
36 maxItems: 100000, minItems: 1
36 minItems: 1, maxItems: 100000
3737
3838 - `custom_id: string`
3939
from line 41
4141
4242 Must be unique for each request within the Message Batch.
4343
44 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,64}$
44 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]{1,64}$
4545
4646 - `params: object`
4747
from line 161
161161
162162 - `document_title: string or null`
163163
164 maxLength: 500, minLength: 1
164 minLength: 1, maxLength: 500
165165
166166 - `end_char_index: number`
167167
from line 181
181181
182182 - `document_title: string or null`
183183
184 maxLength: 500, minLength: 1
184 minLength: 1, maxLength: 500
185185
186186 - `end_page_number: number`
187187
from line 205
205205
206206 - `document_title: string or null`
207207
208 maxLength: 500, minLength: 1
208 minLength: 1, maxLength: 500
209209
210210 - `end_block_index: number`
211211
from line 229
229229
230230 - `title: string or null`
231231
232 maxLength: 512, minLength: 1
232 minLength: 1, maxLength: 512
233233
234234 - `url: string`
235235
from line 385
385385
386386 - `title: optional string or null`
387387
388 maxLength: 500, minLength: 1
388 minLength: 1, maxLength: 500
389389
390390 - `SearchResultBlockParam object`
391391
from line 449
449449
450450 - `name: string`
451451
452 maxLength: 200, minLength: 1
452 minLength: 1, maxLength: 200
453453
454454 - `cache_control: optional CacheControlEphemeral or null`
455455
from line 485
485485
486486 For a toolset member tool_use, the toolset family this member belongs to.
487487
488 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
488 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
489489
490490 - `ToolResultBlockParam object`
491491
from line 521
521521
522522 - `tool_name: string`
523523
524 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
524 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
525525
526526 - `cache_control: optional CacheControlEphemeral or null`
527527
from line 549
549549
550550 The caller-assigned identifier for this tab, unique within the inventory.
551551
552 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
552 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
553553
554554 - `title: string`
555555
from line 575
575575
576576 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
577577
578 maxItems: 200, minItems: 1
578 minItems: 1, maxItems: 200
579579
580580 - `BrowserStateChangeTabOpened object`
581581
from line 593
593593
594594 The `tab_id` of the opened tab, present in `tabs`.
595595
596 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
596 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
597597
598598 - `BrowserStateChangeDownloadStarted object`
599599
from line 605
605605
606606 The caller-assigned identifier for this download, stable across the state changes reporting it.
607607
608 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
608 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
609609
610610 - `url: string`
611611
from line 626
626626
627627 The caller-assigned identifier for this download, stable across the state changes reporting it.
628628
629 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
629 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
630630
631631 - `url: string`
632632
from line 638
638638
639639 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
640640
641 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
641 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
642642
643643 - `size_bytes: optional number or null`
644644
from line 656
656656
657657 The caller-assigned identifier for this download, stable across the state changes reporting it.
658658
659 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
659 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
660660
661661 - `url: string`
662662
from line 668
668668
669669 The failure or cancellation detail, when known.
670670
671 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
671 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
672672
673673 - `is_error: optional boolean`
674674
from line 676
676676
677677 For a toolset member tool_result, the toolset family of the paired tool_use.
678678
679 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
679 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
680680
681681 - `ServerToolUseBlockParam object`
682682
from line 1054
10541054
10551055 - `tool_name: string`
10561056
1057 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
1057 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
10581058
10591059 - `cache_control: optional CacheControlEphemeral or null`
10601060
from line 1095
10951095
10961096 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
10971097
1098 - `string`
1099
10981100 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
10991101
11001102 The model that will complete your prompt.
from line 1139
11371139
11381140 Powerful intelligence for long-running agents and coding
11391141
1140 - `"claude-mythos-preview"`
1141
1142 New class of intelligence, strongest in coding and cybersecurity
1143
11441142 - `"claude-opus-4-6"`
11451143
11461144 Powerful intelligence for long-running agents and coding
from line 1171
11731171
11741172 High-performance model for agents and coding
11751173
1176 - `string`
1174 - `"claude-mythos-preview"`
11771175
1176 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
1177
1178 New class of intelligence, strongest in coding and cybersecurity
1179
11781180 - `cache_control: optional CacheControlEphemeral or null`
11791181
11801182 Top-level cache control automatically applies a cache_control marker to the last cacheable block in the request.
from line 1211
12091211
12101212 Skill ID
12111213
1212 maxLength: 64, minLength: 1
1214 minLength: 1, maxLength: 64
12131215
12141216 - `version: optional string`
12151217
12161218 Skill version or 'latest' for most recent version
12171219
1218 maxLength: 64, minLength: 1
1220 minLength: 1, maxLength: 64
12191221
12201222 - `string`
12211223
12221224 - `diagnostics: optional DiagnosticsParam or null`
12231225
1224 Request-level diagnostics. Currently carries the previous response
1225 id for prompt-cache divergence reporting.
1226 Request-level diagnostics. Supply `previous_message_id` to have the response include `diagnostics.cache_miss_reason` explaining any prompt-cache divergence from that prior request.
12261227
12271228 - `previous_message_id: optional string or null`
12281229
from line 1253
12521253
12531254 - `effort: optional "low" or "medium" or "high" or 2 more or null`
12541255
1255 All possible effort levels.
1256 How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
12561257
1258 Valid values are `low`, `medium`, `high`, `xhigh`, or `max`.
1259
12571260 - `"low"`
12581261
12591262 - `"medium"`
from line 1505
15021505
15031506 This is how the tool will be called by the model and in `tool_use` blocks.
15041507
1505 maxLength: 128, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,128}$
1508 minLength: 1, maxLength: 128, pattern: ^[a-zA-Z0-9_-]{1,128}$
15061509
15071510 - `allowed_callers: optional array of "direct" or "code_execution_20250825" or "code_execution_20260120" or "code_execution_20260521"`
15081511
from line 1722
17191722
17201723 - `configs: optional BrowserToolsetConfigs or null`
17211724
1722 Per-member configuration for `browser_toolset_20260801`: one
1723 optional field per member tool, keyed by the member name — the same
1724 name the member's `tool_use` blocks carry. Every member is an
1725 accepted key, and a member's defaults apply wherever its key is
1726 absent. Unknown keys are rejected: the field set is this toolset
1727 version's complete member set.
1725 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
17281726
17291727 - `type: optional BrowserTypeConfig or null`
17301728
from line 2149
21512149
21522150 - `configs: optional ComputerToolsetConfigs or null`
21532151
2154 Per-member configuration for `computer_toolset_20260801`: one
2155 optional field per member tool, keyed by the member name — the same
2156 name the member's `tool_use` blocks carry. Every member is an
2157 accepted key, and a member's defaults apply wherever its key is
2158 absent. Unknown keys are rejected: the field set is this toolset
2159 version's complete member set.
2152 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
21602153
21612154 - `type: optional ComputerTypeConfig or null`
21622155
from line 2503
25102503
25112504 Maximum number of times the tool can be used in the API request.
25122505
2513 exclusiveMinimum: 0
2506 minimum: 1
25142507
25152508 - `strict: optional boolean`
25162509
from line 2519
25262519
25272520 The city of the user.
25282521
2529 maxLength: 255, minLength: 1
2522 minLength: 1, maxLength: 255
25302523
25312524 - `country: optional string or null`
25322525
25332526 The two letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) of the user.
25342527
2535 maxLength: 2, minLength: 2
2528 minLength: 2, maxLength: 2
25362529
25372530 - `region: optional string or null`
25382531
25392532 The region of the user.
25402533
2541 maxLength: 255, minLength: 1
2534 minLength: 1, maxLength: 255
25422535
25432536 - `timezone: optional string or null`
25442537
25452538 The [IANA timezone](https://nodatime.org/TimeZones) of the user.
25462539
2547 maxLength: 255, minLength: 1
2540 minLength: 1, maxLength: 255
25482541
25492542 - `WebFetchTool20250910 object`
25502543
from line 2583
25902583
25912584 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
25922585
2593 exclusiveMinimum: 0
2586 minimum: 1
25942587
25952588 - `max_uses: optional number or null`
25962589
25972590 Maximum number of times the tool can be used in the API request.
25982591
2599 exclusiveMinimum: 0
2592 minimum: 1
26002593
26012594 - `strict: optional boolean`
26022595
from line 2597
26042597
26052598 - `url_sources: optional WebFetchURLSources or null`
26062599
2607 Which sources contribute to the set of URLs web fetch may fetch.
2600 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
26082601
2609 Each key is a tagged variant: `user_input` is `all` or `none`; the
2610 two tool filters are `all`, `none`, `only` (only the named tools'
2611 results) or `except` (every result but the named tools'). A named tool
2612 must be declared in this request's `tools[]`.
2613
26142602 - `client_tool_results: optional WebFetchURLSourceAll or WebFetchURLSourceNone or WebFetchURLSourceOnly or WebFetchURLSourceExcept`
26152603
26162604 Which client tools' results contribute fetchable URLs: "all", "none", or an only or except list of client tool names from tools[].
from line 2721
27332721
27342722 Maximum number of times the tool can be used in the API request.
27352723
2736 exclusiveMinimum: 0
2724 minimum: 1
27372725
27382726 - `strict: optional boolean`
27392727
from line 2775
27872775
27882776 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
27892777
2790 exclusiveMinimum: 0
2778 minimum: 1
27912779
27922780 - `max_uses: optional number or null`
27932781
27942782 Maximum number of times the tool can be used in the API request.
27952783
2796 exclusiveMinimum: 0
2784 minimum: 1
27972785
27982786 - `strict: optional boolean`
27992787
from line 2789
28012789
28022790 - `url_sources: optional WebFetchURLSources or null`
28032791
2804 Which sources contribute to the set of URLs web fetch may fetch.
2792 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28052793
2806 Each key is a tagged variant: `user_input` is `all` or `none`; the
2807 two tool filters are `all`, `none`, `only` (only the named tools'
2808 results) or `except` (every result but the named tools'). A named tool
2809 must be declared in this request's `tools[]`.
2810
28112794 - `WebFetchTool20260309 object`
28122795
28132796 Web fetch tool with use_cache parameter for bypassing cached content.
from line 2837
28542837
28552838 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
28562839
2857 exclusiveMinimum: 0
2840 minimum: 1
28582841
28592842 - `max_uses: optional number or null`
28602843
28612844 Maximum number of times the tool can be used in the API request.
28622845
2863 exclusiveMinimum: 0
2846 minimum: 1
28642847
28652848 - `strict: optional boolean`
28662849
from line 2851
28682851
28692852 - `url_sources: optional WebFetchURLSources or null`
28702853
2871 Which sources contribute to the set of URLs web fetch may fetch.
2854 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28722855
2873 Each key is a tagged variant: `user_input` is `all` or `none`; the
2874 two tool filters are `all`, `none`, `only` (only the named tools'
2875 results) or `except` (every result but the named tools'). A named tool
2876 must be declared in this request's `tools[]`.
2877
28782856 - `use_cache: optional boolean`
28792857
28802858 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 2897
29192897
29202898 Maximum number of times the tool can be used in the API request.
29212899
2922 exclusiveMinimum: 0
2900 minimum: 1
29232901
29242902 - `response_inclusion: optional "full" or "excluded"`
29252903
from line 2959
29812959
29822960 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
29832961
2984 exclusiveMinimum: 0
2962 minimum: 1
29852963
29862964 - `max_uses: optional number or null`
29872965
29882966 Maximum number of times the tool can be used in the API request.
29892967
2990 exclusiveMinimum: 0
2968 minimum: 1
29912969
29922970 - `response_inclusion: optional "full" or "excluded"`
29932971
from line 2981
30032981
30042982 - `url_sources: optional WebFetchURLSources or null`
30052983
3006 Which sources contribute to the set of URLs web fetch may fetch.
2984 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
30072985
3008 Each key is a tagged variant: `user_input` is `all` or `none`; the
3009 two tool filters are `all`, `none`, `only` (only the named tools'
3010 results) or `except` (every result but the named tools'). A named tool
3011 must be declared in this request's `tools[]`.
3012
30132986 - `use_cache: optional boolean`
30142987
30152988 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 3069
30963069
30973070 Note that even with `temperature` of `0.0`, the results will not be fully deterministic.
30983071
3099 maximum: 1, minimum: 0
3072 minimum: 0, maximum: 1
31003073
31013074 - `top_k: optional number`
31023075
from line 3093
31203093
31213094 Recommended for advanced use cases only.
31223095
3123 maximum: 1, minimum: 0
3096 minimum: 0, maximum: 1
31243097
31253098### Returns
31263099
from line 3441
34683441
34693442 Defaults to `20`. Ranges from `1` to `1000`.
34703443
3471 default: 20, maximum: 1000, minimum: 1
3444 default: 20, minimum: 1, maximum: 1000
34723445
34733446### Headers
34743447
from line 3901
39283901
39293902 - `container: Container or null`
39303903
3931 Information about the container used in the request (for the code execution tool)
3904 Information about the container used in this request.
39323905
3906 This will be non-null if a container tool (e.g. code execution) was used.
3907
39333908 - `id: string`
39343909
39353910 Identifier for the container used in this request
from line 3931
39563931
39573932 Skill ID
39583933
3959 maxLength: 64, minLength: 1
3934 minLength: 1, maxLength: 64
39603935
39613936 - `version: string`
39623937
39633938 The resolved version: a skill version ID for custom skills.
39643939
3965 maxLength: 64, minLength: 1
3940 minLength: 1, maxLength: 64
39663941
39673942 - `content: array of ContentBlock`
39683943
from line 4110
41354110
41364111 - `text: string`
41374112
4138 minLength: 0
4139
41404113 - `ThinkingBlock object`
41414114
41424115 - `type: "thinking"`
from line 4190
42174190
42184191 For a toolset member tool_use, the toolset family.
42194192
4220 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
4193 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
42214194
42224195 - `ServerToolUseBlock object`
42234196
from line 4638
46654638
46664639 - `tool_name: string`
46674640
4668 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
4641 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
46694642
46704643 - `tool_use_id: string`
46714644
from line 4656
46834656
46844657 - `diagnostics: Diagnostics or null`
46854658
4686 Request-level diagnostics: why the prompt cache could not fully reuse
4687 the prefix of the request named by `diagnostics.previous_message_id`.
4659 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
46884660
46894661 - `cache_miss_reason: CacheMissReason or null`
46904662
from line 4720
47484720
47494721 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
47504722
4723 - `string`
4724
47514725 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
47524726
47534727 The model that will complete your prompt.
from line 4764
47904764
47914765 Powerful intelligence for long-running agents and coding
47924766
4793 - `"claude-mythos-preview"`
4794
4795 New class of intelligence, strongest in coding and cybersecurity
4796
47974767 - `"claude-opus-4-6"`
47984768
47994769 Powerful intelligence for long-running agents and coding
from line 4796
48264796
48274797 High-performance model for agents and coding
48284798
4829 - `string`
4799 - `"claude-mythos-preview"`
48304800
4801 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
4802
4803 New class of intelligence, strongest in coding and cybersecurity
4804
48314805 - `role: "assistant"`
48324806
48334807 Conversational role of the generated message.
from line 4812
48384812
48394813 - `stop_details: RefusalStopDetails or null`
48404814
4841 Structured information about a refusal.
4815 Structured information about why model output stopped.
48424816
4817 This is `null` when the `stop_reason` has no additional detail to report.
4818
48434819 - `type: "refusal"`
48444820
48454821 default: refusal
from line 4822
48464822
48474823 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
48484824
4849 The policy category that triggered a refusal.
4825 The policy category that triggered the refusal.
48504826
4827 `null` when the refusal doesn't map to a named category.
4828
48514829 - `"cyber"`
48524830
48534831 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 5412
54345412
54355413 - `container: Container or null`
54365414
5437 Information about the container used in the request (for the code execution tool)
5415 Information about the container used in this request.
54385416
5417 This will be non-null if a container tool (e.g. code execution) was used.
5418
54395419 - `id: string`
54405420
54415421 Identifier for the container used in this request
from line 5442
54625442
54635443 Skill ID
54645444
5465 maxLength: 64, minLength: 1
5445 minLength: 1, maxLength: 64
54665446
54675447 - `version: string`
54685448
54695449 The resolved version: a skill version ID for custom skills.
54705450
5471 maxLength: 64, minLength: 1
5451 minLength: 1, maxLength: 64
54725452
54735453 - `content: array of ContentBlock`
54745454
from line 5621
56415621
56425622 - `text: string`
56435623
5644 minLength: 0
5645
56465624 - `ThinkingBlock object`
56475625
56485626 - `type: "thinking"`
from line 5701
57235701
57245702 For a toolset member tool_use, the toolset family.
57255703
5726 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
5704 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
57275705
57285706 - `ServerToolUseBlock object`
57295707
from line 6149
61716149
61726150 - `tool_name: string`
61736151
6174 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
6152 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
61756153
61766154 - `tool_use_id: string`
61776155
from line 6167
61896167
61906168 - `diagnostics: Diagnostics or null`
61916169
6192 Request-level diagnostics: why the prompt cache could not fully reuse
6193 the prefix of the request named by `diagnostics.previous_message_id`.
6170 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
61946171
61956172 - `cache_miss_reason: CacheMissReason or null`
61966173
from line 6231
62546231
62556232 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
62566233
6234 - `string`
6235
62576236 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
62586237
62596238 The model that will complete your prompt.
from line 6275
62966275
62976276 Powerful intelligence for long-running agents and coding
62986277
6299 - `"claude-mythos-preview"`
6300
6301 New class of intelligence, strongest in coding and cybersecurity
6302
63036278 - `"claude-opus-4-6"`
63046279
63056280 Powerful intelligence for long-running agents and coding
from line 6307
63326307
63336308 High-performance model for agents and coding
63346309
6335 - `string`
6310 - `"claude-mythos-preview"`
63366311
6312 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
6313
6314 New class of intelligence, strongest in coding and cybersecurity
6315
63376316 - `role: "assistant"`
63386317
63396318 Conversational role of the generated message.
from line 6323
63446323
63456324 - `stop_details: RefusalStopDetails or null`
63466325
6347 Structured information about a refusal.
6326 Structured information about why model output stopped.
63486327
6328 This is `null` when the `stop_reason` has no additional detail to report.
6329
63496330 - `type: "refusal"`
63506331
63516332 default: refusal
from line 6333
63526333
63536334 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
63546335
6355 The policy category that triggered a refusal.
6336 The policy category that triggered the refusal.
63566337
6338 `null` when the refusal doesn't map to a named category.
6339
63576340 - `"cyber"`
63586341
63596342 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 6695
67126695
67136696 - `container: Container or null`
67146697
6715 Information about the container used in the request (for the code execution tool)
6698 Information about the container used in this request.
67166699
6700 This will be non-null if a container tool (e.g. code execution) was used.
6701
67176702 - `id: string`
67186703
67196704 Identifier for the container used in this request
from line 6725
67406725
67416726 Skill ID
67426727
6743 maxLength: 64, minLength: 1
6728 minLength: 1, maxLength: 64
67446729
67456730 - `version: string`
67466731
67476732 The resolved version: a skill version ID for custom skills.
67486733
6749 maxLength: 64, minLength: 1
6734 minLength: 1, maxLength: 64
67506735
67516736 - `content: array of ContentBlock`
67526737
from line 6904
69196904
69206905 - `text: string`
69216906
6922 minLength: 0
6923
69246907 - `ThinkingBlock object`
69256908
69266909 - `type: "thinking"`
from line 6984
70016984
70026985 For a toolset member tool_use, the toolset family.
70036986
7004 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
6987 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
70056988
70066989 - `ServerToolUseBlock object`
70076990
from line 7432
74497432
74507433 - `tool_name: string`
74517434
7452 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
7435 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
74537436
74547437 - `tool_use_id: string`
74557438
from line 7450
74677450
74687451 - `diagnostics: Diagnostics or null`
74697452
7470 Request-level diagnostics: why the prompt cache could not fully reuse
7471 the prefix of the request named by `diagnostics.previous_message_id`.
7453 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
74727454
74737455 - `cache_miss_reason: CacheMissReason or null`
74747456
from line 7514
75327514
75337515 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
75347516
7517 - `string`
7518
75357519 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
75367520
75377521 The model that will complete your prompt.
from line 7558
75747558
75757559 Powerful intelligence for long-running agents and coding
75767560
7577 - `"claude-mythos-preview"`
7578
7579 New class of intelligence, strongest in coding and cybersecurity
7580
75817561 - `"claude-opus-4-6"`
75827562
75837563 Powerful intelligence for long-running agents and coding
from line 7590
76107590
76117591 High-performance model for agents and coding
76127592
7613 - `string`
7593 - `"claude-mythos-preview"`
76147594
7595 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
7596
7597 New class of intelligence, strongest in coding and cybersecurity
7598
76157599 - `role: "assistant"`
76167600
76177601 Conversational role of the generated message.
from line 7606
76227606
76237607 - `stop_details: RefusalStopDetails or null`
76247608
7625 Structured information about a refusal.
7609 Structured information about why model output stopped.
76267610
7611 This is `null` when the `stop_reason` has no additional detail to report.
7612
76277613 - `type: "refusal"`
76287614
76297615 default: refusal
from line 7616
76307616
76317617 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
76327618
7633 The policy category that triggered a refusal.
7619 The policy category that triggered the refusal.
76347620
7621 `null` when the refusal doesn't map to a named category.
7622
76357623 - `"cyber"`
76367624
76377625 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 7930
79427930
79437931 - `container: Container or null`
79447932
7945 Information about the container used in the request (for the code execution tool)
7933 Information about the container used in this request.
79467934
7935 This will be non-null if a container tool (e.g. code execution) was used.
7936
79477937 - `id: string`
79487938
79497939 Identifier for the container used in this request
from line 7960
79707960
79717961 Skill ID
79727962
7973 maxLength: 64, minLength: 1
7963 minLength: 1, maxLength: 64
79747964
79757965 - `version: string`
79767966
79777967 The resolved version: a skill version ID for custom skills.
79787968
7979 maxLength: 64, minLength: 1
7969 minLength: 1, maxLength: 64
79807970
79817971 - `content: array of ContentBlock`
79827972
from line 8139
81498139
81508140 - `text: string`
81518141
8152 minLength: 0
8153
81548142 - `ThinkingBlock object`
81558143
81568144 - `type: "thinking"`
from line 8219
82318219
82328220 For a toolset member tool_use, the toolset family.
82338221
8234 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
8222 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
82358223
82368224 - `ServerToolUseBlock object`
82378225
from line 8667
86798667
86808668 - `tool_name: string`
86818669
8682 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
8670 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
86838671
86848672 - `tool_use_id: string`
86858673
from line 8685
86978685
86988686 - `diagnostics: Diagnostics or null`
86998687
8700 Request-level diagnostics: why the prompt cache could not fully reuse
8701 the prefix of the request named by `diagnostics.previous_message_id`.
8688 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
87028689
87038690 - `cache_miss_reason: CacheMissReason or null`
87048691
from line 8721
87348721
87358722 Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.
87368723
8737 - `
8724 - `CacheMissMessagesChanged object`
8725
8726 - `type
No line in this hunk matches that.