@ai-sdk/anthropic
4.0.106 · 17 Nov 2026Apache-2.024.5M / weekRepositoryHomepagenpmAtom feed
457 stable releases, newest first, by release line. Notes are quoted as the maintainers published them.
4.0.106
17 Nov 2026Only updates dependencies:@ai-sdk/provider 4.0.39@ai-sdk/provider-utils 5.0.82
4.0.105
13 Nov 2026Derive known model ID suggestions from each provider's model catalog declarations so catalog updates automatically update autocomplete, while preserving existing suggestions, endpoint-specific model lists, and support for custom model IDs.
Patch Changes
Derive known model ID suggestions from each provider's model catalog declarations so catalog updates automatically update autocomplete, while preserving existing suggestions, endpoint-specific model lists, and support for custom model IDs.
4.0.104
13 Nov 2026Add experimental extraction models backed by structured language model output
Patch Changes
Add experimental extraction models backed by structured language model output
4.0.103
12 Nov 2026Only updates dependencies:@ai-sdk/provider-utils 5.0.80@ai-sdk/provider 4.0.38
4.0.102
10 Nov 2026Add MiniMax H3 Image and H3 Image Highspeed support for text-to-image generation and image editing, including output controls and token usage details.
Patch Changes
Add MiniMax H3 Image and H3 Image Highspeed support for text-to-image generation and image editing, including output controls and token usage details.
4.0.101
10 Nov 2026Only updates a dependency:@ai-sdk/provider-utils 5.0.78
4.0.99
7 Nov 2026Keep computer toolset actions tied to the returned member name and reject disabled or unavailable members before emitting tool calls. Apply member configuration to local input validation and reject malformed or non-object member input.
Add
resolveProviderOptionsto parse provider options that a model accepts under severalproviderOptionskeys, either as aliases (first) or merged in order (merge), and use it for the existing multi-key lookups in the OpenAI, OpenAI-compatible, Anthropic, Google, and Amazon Bedrock models. Resolution behavior for these models is unchanged.Reject conflicting inline tool additions when a named tool choice falls back to auto or the internal JSON response tool limits available application tools, including definitions replayed from conversation history.
Patch Changes
Keep computer toolset actions tied to the returned member name and reject disabled or unavailable members before emitting tool calls. Apply member configuration to local input validation and reject malformed or non-object member input.
Add
resolveProviderOptionsto parse provider options that a model accepts under severalproviderOptionskeys, either as aliases (first) or merged in order (merge), and use it for the existing multi-key lookups in the OpenAI, OpenAI-compatible, Anthropic, Google, and Amazon Bedrock models. Resolution behavior for these models is unchanged.Reject conflicting inline tool additions when a named tool choice falls back to auto or the internal JSON response tool limits available application tools, including definitions replayed from conversation history.
Disable MCP connector tools when structured output falls back to the JSON response tool and the caller selects no tools or a named application tool, and when a named application tool choice falls back to automatic tool selection.
Preserve warning-only handling of ignored inline tool additions on the initial system message when named tool choice falls back or the JSON response tool is used. Validate tool-choice restrictions only for inline definitions emitted in the outgoing request.
4.0.98
6 Nov 2026Refactor Amazon Bedrock model ID logic into a model catalog that declares capabilities and provider options, and generate provider capability documentation from it. Export the Anthropic catalog capability helper under its compatibility name for Bedrock to reuse. Request behavior is unchanged.
Refactor Anthropic AWS model ID logic into a model catalog that declares capabilities and provider options, and generate provider capability documentation from it. Export Anthropic catalog metadata internally for Anthropic AWS to reuse. Request behavior is unchanged.
anthropic: Support adding tools by value in mid-conversation
toolChanges(inline-tools-2026-09-15)
Patch Changes
Refactor Amazon Bedrock model ID logic into a model catalog that declares capabilities and provider options, and generate provider capability documentation from it. Export the Anthropic catalog capability helper under its compatibility name for Bedrock to reuse. Request behavior is unchanged.
Refactor Anthropic AWS model ID logic into a model catalog that declares capabilities and provider options, and generate provider capability documentation from it. Export Anthropic catalog metadata internally for Anthropic AWS to reuse. Request behavior is unchanged.
anthropic: Support adding tools by value in mid-conversation
toolChanges(inline-tools-2026-09-15)Add support for creating and continuing Anthropic message threads, including automatic beta headers and thread error details for application-managed recovery.
4.0.97
5 Nov 2026Preserve the order of text and complete function calls in streamed Google responses, including standalone thought signatures on empty text or reasoning parts.
Keep provider metadata on empty text parts when converting results to response messages so it can be passed back in subsequent requests.
Omit foreign-provider empty signature text when replaying history through Anthropic or Bedrock, preserving Bedrock’s existing native reasoning-history handling.
Add MiniMax
serviceTiersupport forstandardandpriorityadmission. Omit tiers belonging to another known provider with an unsupported-setting warning instead of forwarding incompatible values. Expose the appliedserviceTier, when returned, in Anthropic-compatible generation, stream and batch metadata without changing existing metadata construction types.Add typed streaming heartbeat parts. Anthropic emits them only with
includePings: true(disabled by default), without requiring raw chunks. Core forwards pings through stream retries without storing them as generated content or sending them to UI message streams. Consumers with exhaustive stream-part switches must handle the newpingunion variant when upgrading, even when the opt-in is disabled. Typed pings do not flush partial words during stream smoothing.
Patch Changes
Preserve the order of text and complete function calls in streamed Google responses, including standalone thought signatures on empty text or reasoning parts.
Keep provider metadata on empty text parts when converting results to response messages so it can be passed back in subsequent requests.
Omit foreign-provider empty signature text when replaying history through Anthropic or Bedrock, preserving Bedrock’s existing native reasoning-history handling.
Add MiniMax
serviceTiersupport forstandardandpriorityadmission. Omit tiers belonging to another known provider with an unsupported-setting warning instead of forwarding incompatible values. Expose the appliedserviceTier, when returned, in Anthropic-compatible generation, stream and batch metadata without changing existing metadata construction types.Add typed streaming heartbeat parts. Anthropic emits them only with
includePings: true(disabled by default), without requiring raw chunks. Core forwards pings through stream retries without storing them as generated content or sending them to UI message streams. Consumers with exhaustive stream-part switches must handle the newpingunion variant when upgrading, even when the opt-in is disabled. Typed pings do not flush partial words during stream smoothing.Add a
supportsNativeStructuredOutputopt-in to the Vertex Anthropic provider soautocan use native structured output for Claude 4.5 and newer when enabled by Google Cloud policy. Keep tool emulation as the default and avoid adding the structured-output beta header for providers without strict tool support.Support deleting files, retrieving metadata, and streaming downloads of generated files through the Anthropic Files API. Metadata reads and uploads expose consistent media type, size, creation time, expiration, and provider metadata when available. Downloads preserve Anthropic’s restriction to files generated by tools or skills.
4.0.96
3 Nov 2026Only updates a dependency:@ai-sdk/provider-utils 5.0.74
4.0.95
2 Nov 2026Enable native structured output and strict tools for supported Claude 5 models on Bedrock. In automatic mode, Bedrock and Anthropic now fall back for schemas that native structured output cannot represent: recursive references, open objects, or more than ten optional properties. Anthropic uses the JSON response tool; Bedrock Converse uses the JSON tool or JSON instructions for models that reject forced tool use. The JSON-tool fallback excludes disabled and unselected application tools. Explicit
outputFormatselection remains unchanged.anthropic: Parse valid JSON string tool input and safely serialize BigInt and cycles in tool results
Patch Changes
Enable native structured output and strict tools for supported Claude 5 models on Bedrock. In automatic mode, Bedrock and Anthropic now fall back for schemas that native structured output cannot represent: recursive references, open objects, or more than ten optional properties. Anthropic uses the JSON response tool; Bedrock Converse uses the JSON tool or JSON instructions for models that reject forced tool use. The JSON-tool fallback excludes disabled and unselected application tools. Explicit
outputFormatselection remains unchanged.anthropic: Parse valid JSON string tool input and safely serialize BigInt and cycles in tool results
4.0.94
30 Oct 2026anthropic: Use native structured outputs for Claude models that reject forced tool use (Sonnet 5.5, Opus 5.5, Fable 5.1) even when the provider disables them by default, as Amazon Bedrock and Google Vertex do. Self-referencing schemas keep the JSON response tool. Also keep
definitionsand$defsnext to a top-level$refwhen sanitizing structured output schemas, so the reference still resolves.anthropic: When the JSON response tool is in use but the model answers in text without calling it (for example with manual thinking, where the forced tool choice is downgraded to
auto), return the final answer text instead of dropping it. Text before a tool call, result, or message continuation is still dropped.anthropic: Keep the JSON response tool for schemas that native structured outputs cannot represent when upgrading models that reject forced tool use. Schemas with open objects (maps, free-form objects) would otherwise come back as
{}, and schemas with more than 10 optional properties can be rejected as too complex.
4.0.93
30 Oct 2026Introduce a model entry contract with named capability sections and request configuration. Consolidate the OpenAI and Anthropic model catalogs so existing request behavior and generated capability documentation use the same source of truth. Catalogs also declare the provider options each provider accepts and which models support them.
Patch Changes
Introduce a model entry contract with named capability sections and request configuration. Consolidate the OpenAI and Anthropic model catalogs so existing request behavior and generated capability documentation use the same source of truth. Catalogs also declare the provider options each provider accepts and which models support them.
4.0.92
29 Oct 2026Map the top-level
reasoningoption consistently across providersProviders now resolve reasoning the same way:
mapReasoningToProviderEffortnow accepts anyreasoningvalue, returns an optionalproviderOptionsEffortunchanged, and rounds levels that are missing from the effort map to the closest supported level.mapReasoningToProviderBudgetmaps'none'to the minimum reasoning budget, always returns a budget, and uses the default percentage for levels missing frombudgetPercentages.OpenAICompatibleChatConfigadds an optionalmapReasoningfunction for mapping the top-levelreasoningoption.
Patch Changes
Map the top-level
reasoningoption consistently across providersProviders now resolve reasoning the same way:
A provider-specific reasoning setting in
providerOptionsis sent unchanged. OpenAI no longer dropsproviderOptions.openai.reasoningEffortvalues that are not in its list for the model, and Fireworks no longer rewritesproviderOptions.fireworks.reasoningEffort. Fireworks also sends top-levelreasoningvalues directly instead of rounding them to'high'at most, except for model families known to reject'none'(GPT-OSS, GLM-5.3, Muse Spark) or'minimal'(Nemotron, Inkling).Otherwise,
reasoningmaps to the closest level the model supports, rounding down, and rounds up only when there is no lower level. For example,'xhigh'now maps to'high'instead of'max'on Anthropic and Bedrock models withoutxhigheffort, and'medium'maps to'low'instead of'high'on DeepSeek, Kimi K3, and Mistral GLM-5.3 models.'none'on a model that cannot disable reasoning maps to the model's lowest level with a compatibility warning, instead of being dropped or sending a request the API rejects. This affects GPT-6 models without'none', xAI grok-4.5 and later, Groq GPT-OSS, Kimi K3, Perplexity, QuiverAI, Bedrock OpenAIgpt-ossmodels, and Gemini 2.5 Pro (minimum thinking budget). Gemini 3 and Anthropic and Bedrockbetween_toolsmappings now emit the compatibility warning too.
mapReasoningToProviderEffortnow accepts anyreasoningvalue, returns an optionalproviderOptionsEffortunchanged, and rounds levels that are missing from the effort map to the closest supported level.mapReasoningToProviderBudgetmaps'none'to the minimum reasoning budget, always returns a budget, and uses the default percentage for levels missing frombudgetPercentages.OpenAICompatibleChatConfigadds an optionalmapReasoningfunction for mapping the top-levelreasoningoption.
4.0.91
28 Oct 2026Fix workflow serialization of Anthropic models using API keys or synchronous auth tokens while preserving support for asynchronous token providers.
4.0.90
27 Oct 2026Support synchronous and asynchronous
authTokenproviders so applications can use refreshable credentials, including Anthropic workload identity federation helpers for inference. Resolve tokens before each request while preserving existing authentication headers and beta-header behavior.
Patch Changes
Support synchronous and asynchronous
authTokenproviders so applications can use refreshable credentials, including Anthropic workload identity federation helpers for inference. Resolve tokens before each request while preserving existing authentication headers and beta-header behavior.
4.0.87 – 4.0.893 releases that only update dependencies@ai-sdk/provider-utils 5.0.69@ai-sdk/provider 4.0.34Show each
4.0.86
22 Oct 2026Report
interruptedwhen a stream ends cleanly without its provider completion signal, preserving partial output without emitting an error solely for the interruption.This applies to OpenAI (Chat, Responses, and Completions), Anthropic, Google (Generate Content and Interactions), Alibaba, Amazon Bedrock Converse, Cohere, DeepSeek, Groq, Hugging Face Responses, Mistral Chat, Moonshot AI, Open Responses, Perplexity, and xAI Responses. It also covers cutoffs in later Anthropic messages and Google automatic continuations.
Compatibility notice: These streams previously fell back to
otheror omitted the finish part. Applications must now handleinterruptedin finish-reason switches, mappings, and custom validation. CheckfinishReason === 'interrupted'to detect incomplete responses; interruption alone does not trigger error handling.Explicit unknown completion reasons still map to
other. Actual provider, parsing, and transport errors retain their error paths; affected error paths that previously left the finish reason asothernow reporterror. OpenAI-compatible remains unchanged.anthropic: Preserve empty-stream rejection for interrupted responses
Patch Changes
Report
interruptedwhen a stream ends cleanly without its provider completion signal, preserving partial output without emitting an error solely for the interruption.This applies to OpenAI (Chat, Responses, and Completions), Anthropic, Google (Generate Content and Interactions), Alibaba, Amazon Bedrock Converse, Cohere, DeepSeek, Groq, Hugging Face Responses, Mistral Chat, Moonshot AI, Open Responses, Perplexity, and xAI Responses. It also covers cutoffs in later Anthropic messages and Google automatic continuations.
Compatibility notice: These streams previously fell back to
otheror omitted the finish part. Applications must now handleinterruptedin finish-reason switches, mappings, and custom validation. CheckfinishReason === 'interrupted'to detect incomplete responses; interruption alone does not trigger error handling.Explicit unknown completion reasons still map to
other. Actual provider, parsing, and transport errors retain their error paths; affected error paths that previously left the finish reason asothernow reporterror. OpenAI-compatible remains unchanged.anthropic: Preserve empty-stream rejection for interrupted responses
4.0.85
21 Oct 2026Add a shared encodePathSegment helper and use it to encode provider-returned identifiers in credentialed follow-up request paths.
Defer construction of provider validation schemas until first use to reduce package import time.
Defer construction of Anthropic provider option schemas until they are first used to reduce package import time.
Patch Changes
Add a shared encodePathSegment helper and use it to encode provider-returned identifiers in credentialed follow-up request paths.
Defer construction of provider validation schemas until first use to reduce package import time.
Defer construction of Anthropic provider option schemas until they are first used to reduce package import time.
4.0.84
20 Oct 2026Add the MiniMax M3.1 Flash model ID to the MiniMax and AI Gateway providers.
Patch Changes
Add the MiniMax M3.1 Flash model ID to the MiniMax and AI Gateway providers.
4.0.82 – 4.0.832 releases that only update dependencies@ai-sdk/provider 4.0.30@ai-sdk/provider-utils 5.0.63Show each
4.0.81
15 Oct 2026anthropic: Support provider-referenced files in tool results
Patch Changes
anthropic: Support provider-referenced files in tool results
4.0.80
15 Oct 2026vertex-anthropic: Forward 'providerOptions.vertex.labels' as a header
Fall back to automatic tool choice with a warning when manual thinking is enabled with a required or named tool choice, avoiding an Anthropic API 400 error. Named tool choices send only the selected tool.
Patch Changes
vertex-anthropic: Forward 'providerOptions.vertex.labels' as a header
Fall back to automatic tool choice with a warning when manual thinking is enabled with a required or named tool choice, avoiding an Anthropic API 400 error. Named tool choices send only the selected tool.
4.0.79
12 Oct 2026anthropic: Preserve tools when disabling tool calls
anthropic: Preserve failed tool search results in conversation history
Keep retrieved web-search sources separate from normalized inline citations on text parts and text-end events. Preserve every supported web and document citation kind in generation, streaming, and batch results, with typed AnthropicCitation and AnthropicTextProviderMetadata exports. Retain document source compatibility and web citation source fallbacks when retrieval results are unavailable.
Patch Changes
anthropic: Preserve tools when disabling tool calls
anthropic: Preserve failed tool search results in conversation history
Keep retrieved web-search sources separate from normalized inline citations on text parts and text-end events. Preserve every supported web and document citation kind in generation, streaming, and batch results, with typed AnthropicCitation and AnthropicTextProviderMetadata exports. Retain document source compatibility and web citation source fallbacks when retrieval results are unavailable.
4.0.78
8 Oct 2026Only updates dependencies:@ai-sdk/provider 4.0.26@ai-sdk/provider-utils 5.0.58
4.0.77
8 Oct 2026anthropic: Avoid unsupported strict mode warnings when tools set
strict: false
4.0.76
8 Oct 2026Only updates dependencies:@ai-sdk/provider 4.0.25@ai-sdk/provider-utils 5.0.57
4.0.75
7 Oct 2026anthropic: Add Claude Haiku 5.5 support
4.0.74
7 Oct 2026Only updates dependencies:@ai-sdk/provider 4.0.24@ai-sdk/provider-utils 5.0.56
4.0.73
6 Oct 2026Limit jsonl rows to 64mb during batch result downloads
provider: Add a portable
maxreasoning level with native provider mappings, compatibility coercions, and budget-based fallback support.Compatibility notes:
Release classification: This remains a patch changeset under the repository's explicit release policy, which uses patch releases for both fixes and features. The TypeScript source-compatibility caveat for exhaustive consumers is disclosed above. Maintainers can override this classification with a
majorlabel if they decide to align the provider-spec addition with a future AI SDK major release.
Patch Changes
Limit jsonl rows to 64mb during batch result downloads
provider: Add a portable
maxreasoning level with native provider mappings, compatibility coercions, and budget-based fallback support.Compatibility notes:
TypeScript source compatibility: Adding
maxwidensLanguageModelV4CallOptions['reasoning']and the derivedReasoningLeveltype. Third-party providers and consumers with exhaustive switches orRecord<ReasoningLevel, ...>mappings must handlemaxbefore upgrading. Ordinary calls using existing reasoning levels remain accepted.Amazon Bedrock: Existing Nova 2 calls using portable
reasoning: 'xhigh'now sendmaxReasoningEffort: 'high'instead of the unsupported'max', and return a compatibility warning. Anthropic and OpenAI Bedrock effort mappings retain their existingxhighbehavior. Partial Bedrock reasoning configurations continue to preserve explicit values and derive missing fields; explicit effort/budget overrides and disabled thinking no longer emit warnings for portable mappings that are not sent.Fireworks: Existing portable
minimalandxhighcalls keep their request mappings tolowandhigh, respectively, and now return compatibility warnings in generation and streaming results. Portablemaxmaps tohighwith the same warning behavior.
Release classification: This remains a patch changeset under the repository's explicit release policy, which uses patch releases for both fixes and features. The TypeScript source-compatibility caveat for exhaustive consumers is disclosed above. Maintainers can override this classification with a
majorlabel if they decide to align the provider-spec addition with a future AI SDK major release.
4.0.72
5 Oct 2026Rename
evaluatetodecideanthropic: Expose message-start usage through custom stream parts
anthropic: Add the required beta when replaying server-side fallback blocks
Patch Changes
Rename
evaluatetodecideanthropic: Expose message-start usage through custom stream parts
anthropic: Add the required beta when replaying server-side fallback blocks
4.0.71
30 Sep 2026anthropic: Preserve invalid toolset calls
4.0.70
30 Sep 2026Only updates dependencies:@ai-sdk/provider 4.0.21@ai-sdk/provider-utils 5.0.53
4.0.69
30 Sep 2026anthropic: Preserve server-side fallback boundaries when replaying assistant messages
Patch Changes
anthropic: Preserve server-side fallback boundaries when replaying assistant messages
4.0.68
28 Sep 2026Use standards-compliant User-Agent header
Patch Changes
Use standards-compliant User-Agent header
4.0.67
28 Sep 2026anthropic: Add Claude Sonnet 5.5 support
Normalize dangling programmatic tool caller references in conversation history after pruning. Before a subsequent user message, omit caller metadata whose source code execution call is missing and emit a warning, preserving the retained tool calls and results. Keep caller metadata unchanged for active tool continuations.
4.0.65 – 4.0.662 releases that only update dependencies@ai-sdk/provider-utils 5.0.50@ai-sdk/provider 4.0.19Show each
4.0.63
24 Sep 2026anthropic: Support on-demand compaction and preserve signed compaction blocks
4.0.62
23 Sep 2026Only updates a dependency:@ai-sdk/provider-utils 5.0.47
4.0.61
23 Sep 2026anthropic: Omit empty compaction blocks from replay
Patch Changes
anthropic: Omit empty compaction blocks from replay
4.0.60
22 Sep 2026anthropic: Add Claude Opus 5.5 support
4.0.59
22 Sep 2026provider/anthropic: Add
safeguardsprovider option andsafeguardResultsprovider metadata (dangerous tool use classifier)
4.0.57 – 4.0.582 releases that only update dependencies@ai-sdk/provider-utils 5.0.45Show each
- 4.0.5818 Sep 2026@ai-sdk/provider-utils 5.0.45
- 4.0.5718 Sep 2026@ai-sdk/provider-utils 5.0.44
4.0.56
16 Sep 2026Add
anthropic.evaluationModel()for experimental Choice, Score, and Boolean evaluations through Messages structured output, with exact labels, validated score bounds, and prompted Boolean P(true) estimates validated to be in [0, 1]. Boolean estimates are not guaranteed to be calibrated; application code chooses thresholds.
Patch Changes
Add
anthropic.evaluationModel()for experimental Choice, Score, and Boolean evaluations through Messages structured output, with exact labels, validated score bounds, and prompted Boolean P(true) estimates validated to be in [0, 1]. Boolean estimates are not guaranteed to be calibrated; application code chooses thresholds.
4.0.55
16 Sep 2026anthropic: Add 20260318 web search and web fetch tools with response inclusion controls
Patch Changes
anthropic: Add 20260318 web search and web fetch tools with response inclusion controls
A whole project to check? Paste its package.json in the dependency check: every deprecated, outdated or vulnerable dependency, with what to upgrade to. Free, no account.