@ai-sdk/amazon-bedrock
5.0.145 · 17 Nov 2026Apache-2.06.9M / weekRepositoryHomepagenpmAtom feed
575 stable releases, newest first, by release line. Notes are quoted as the maintainers published them.
5.0.143 – 5.0.1453 releases that only update dependencies@ai-sdk/openai 4.0.121@ai-sdk/provider 4.0.39@ai-sdk/provider-utils 5.0.82and 1 moreShow each
- 5.0.14517 Nov 2026@ai-sdk/openai 4.0.121
- 5.0.14417 Nov 2026@ai-sdk/openai 4.0.120
- 5.0.14317 Nov 2026@ai-sdk/openai 4.0.119@ai-sdk/provider 4.0.39@ai-sdk/provider-utils 5.0.82and 1 more
5.0.142
13 Nov 2026amazon-bedrock: Send
block_binding.prefix_mismatch_behaviorto Bedrock again instead of the deprecatedmismatch_behavioraliasDerive 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
amazon-bedrock: Send
block_binding.prefix_mismatch_behaviorto Bedrock again instead of the deprecatedmismatch_behavioraliasDerive 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.
5.0.137 – 5.0.1415 releases that only update dependencies@ai-sdk/provider-utils 5.0.81@ai-sdk/openai 4.0.117@ai-sdk/anthropic 4.0.104and 1 moreShow each
- 5.0.14113 Nov 2026@ai-sdk/provider-utils 5.0.81@ai-sdk/openai 4.0.117@ai-sdk/anthropic 4.0.104
- 5.0.14012 Nov 2026@ai-sdk/openai 4.0.116@ai-sdk/provider-utils 5.0.80@ai-sdk/provider 4.0.38and 1 more
- 5.0.13911 Nov 2026@ai-sdk/openai 4.0.115
- 5.0.13810 Nov 2026@ai-sdk/provider-utils 5.0.79@ai-sdk/anthropic 4.0.102@ai-sdk/openai 4.0.114
- 5.0.13710 Nov 2026@ai-sdk/openai 4.0.113@ai-sdk/provider-utils 5.0.78@ai-sdk/anthropic 4.0.101
5.0.135
7 Nov 2026Add
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.
Patch Changes
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.
5.0.134
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.
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.
5.0.133
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.
Pass through
seed: 0instead of dropping it in image and video models. KlingAI now reportsseed: 0as an unsupported setting like any other seed.Report the actual OpenAI Chat HTTP endpoint (also used by Azure Chat and Bedrock Mantle Chat) through optional request metadata and model-call end events. Record its host and port on OpenTelemetry chat spans without exposing URL credentials, paths, queries, or fragments. Expose only the sanitized HTTP origin in text and object generation results and step/finish callbacks. Providers that do not report an endpoint leave these attributes unset.
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.
Pass through
seed: 0instead of dropping it in image and video models. KlingAI now reportsseed: 0as an unsupported setting like any other seed.Report the actual OpenAI Chat HTTP endpoint (also used by Azure Chat and Bedrock Mantle Chat) through optional request metadata and model-call end events. Record its host and port on OpenTelemetry chat spans without exposing URL credentials, paths, queries, or fragments. Expose only the sanitized HTTP origin in text and object generation results and step/finish callbacks. Providers that do not report an endpoint leave these attributes unset.
5.0.132
3 Nov 2026Only updates dependencies:@ai-sdk/provider-utils 5.0.74@ai-sdk/anthropic 4.0.96@ai-sdk/openai 4.0.108
5.0.131
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.amazon-bedrock: Keep every function tool on a named tool choice for Anthropic models
Anthropic models that accept forced tool use now receive every function tool on a named tool choice, as with
@ai-sdk/anthropic, so a forced step no longer invalidates the prompt cache. Parallel tool use is disabled on that step by default, so the model still calls only the named tool. SetproviderOptions.anthropic.disableParallelToolUse: falseto allow parallel calls. Other models, and Anthropic models that reject forced tool use, still receive only the named tool.amazon-bedrock: Parse JSON string tool inputs and safely serialize BigInt and cycles in tool outputs
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.amazon-bedrock: Keep every function tool on a named tool choice for Anthropic models
Anthropic models that accept forced tool use now receive every function tool on a named tool choice, as with
@ai-sdk/anthropic, so a forced step no longer invalidates the prompt cache. Parallel tool use is disabled on that step by default, so the model still calls only the named tool. SetproviderOptions.anthropic.disableParallelToolUse: falseto allow parallel calls. Other models, and Anthropic models that reject forced tool use, still receive only the named tool.amazon-bedrock: Parse JSON string tool inputs and safely serialize BigInt and cycles in tool outputs
5.0.130
30 Oct 2026Only updates a dependency:@ai-sdk/anthropic 4.0.94
5.0.129
30 Oct 2026Support promises and synchronous or asynchronous callbacks in provider API key and token settings. Resolve credentials per authenticated request while preserving existing authentication behavior and synchronous credential serialization.
The experimental OpenAI Live
getServerWebSocketConfig()method can now resolve asynchronously. Await its result before using the connection settings.
Patch Changes
Support promises and synchronous or asynchronous callbacks in provider API key and token settings. Resolve credentials per authenticated request while preserving existing authentication behavior and synchronous credential serialization.
The experimental OpenAI Live
getServerWebSocketConfig()method can now resolve asynchronously. Await its result before using the connection settings.
5.0.128
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.
5.0.127
28 Oct 2026Only updates a dependency:@ai-sdk/anthropic 4.0.91
5.0.126
27 Oct 2026amazon-bedrock: Make document names unique within a request
Bedrock Converse rejects requests that contain duplicate document names. Different filenames could sanitize to the same name (e.g.
rates.v1.pdfandrates.v2.pdfboth becamerates), and repeated filenames were sent as-is, so these requests failed. Repeated names now get a numeric suffix such asinvoice (2).Forward disabled thinking in Anthropic Converse requests when using
reasoning: 'none'orreasoningConfig: { type: 'disabled' }, omitting the unsupported thinking field for Claude 3.5 and earlier. Apply model-specific fallbacks and effort limits when disabled thinking is unsupported, preserving explicit reasoning effort during the adaptive fallback.
Patch Changes
amazon-bedrock: Make document names unique within a request
Bedrock Converse rejects requests that contain duplicate document names. Different filenames could sanitize to the same name (e.g.
rates.v1.pdfandrates.v2.pdfboth becamerates), and repeated filenames were sent as-is, so these requests failed. Repeated names now get a numeric suffix such asinvoice (2).Forward disabled thinking in Anthropic Converse requests when using
reasoning: 'none'orreasoningConfig: { type: 'disabled' }, omitting the unsupported thinking field for Claude 3.5 and earlier. Apply model-specific fallbacks and effort limits when disabled thinking is unsupported, preserving explicit reasoning effort during the adaptive fallback.
5.0.125
26 Oct 2026Fix string interpolation in Amazon Bedrock unsupported provider tool warnings and OpenAI / OpenAI-compatible completion errors for unexpected system messages.
Patch Changes
Fix string interpolation in Amazon Bedrock unsupported provider tool warnings and OpenAI / OpenAI-compatible completion errors for unexpected system messages.
5.0.123 – 5.0.1242 releases that only update dependencies@ai-sdk/openai 4.0.102@ai-sdk/provider 4.0.34@ai-sdk/provider-utils 5.0.68and 1 moreShow each
- 5.0.12424 Oct 2026@ai-sdk/openai 4.0.102@ai-sdk/provider 4.0.34@ai-sdk/provider-utils 5.0.68and 1 more
- 5.0.12322 Oct 2026@ai-sdk/provider-utils 5.0.67@ai-sdk/provider 4.0.33@ai-sdk/anthropic 4.0.87and 1 more
5.0.122
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.
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.
5.0.121
21 Oct 2026Defer construction of provider validation schemas until first use to reduce package import time.
Patch Changes
Defer construction of provider validation schemas until first use to reduce package import time.
5.0.115 – 5.0.1206 releases that only update dependencies@ai-sdk/anthropic 4.0.84@ai-sdk/provider 4.0.31@ai-sdk/provider-utils 5.0.64and 1 moreShow each
- 5.0.12020 Oct 2026@ai-sdk/anthropic 4.0.84@ai-sdk/provider 4.0.31@ai-sdk/provider-utils 5.0.64and 1 more
- 5.0.11919 Oct 2026@ai-sdk/openai 4.0.97@ai-sdk/provider 4.0.30@ai-sdk/anthropic 4.0.83and 1 more
- 5.0.11816 Oct 2026@ai-sdk/openai 4.0.96@ai-sdk/provider 4.0.29@ai-sdk/provider-utils 5.0.62and 1 more
- 5.0.11715 Oct 2026@ai-sdk/provider 4.0.28@ai-sdk/openai 4.0.95@ai-sdk/anthropic 4.0.81and 1 more
- 5.0.11615 Oct 2026@ai-sdk/provider-utils 5.0.60@ai-sdk/anthropic 4.0.80@ai-sdk/openai 4.0.94
- 5.0.11513 Oct 2026@ai-sdk/openai 4.0.93
5.0.114
12 Oct 2026Use known supported Claude models for native structured output and strict tools on Bedrock. Future Claude model IDs omit those unsupported fields and use JSON instructions for structured output by default.
Patch Changes
Use known supported Claude models for native structured output and strict tools on Bedrock. Future Claude model IDs omit those unsupported fields and use JSON instructions for structured output by default.
5.0.113
8 Oct 2026Map the top level embedding dimensions to provider settings
Patch Changes
Map the top level embedding dimensions to provider settings
5.0.112
8 Oct 2026anthropic: Avoid unsupported strict mode warnings when tools set
strict: false
Patch Changes
anthropic: Avoid unsupported strict mode warnings when tools set
strict: false
5.0.111
8 Oct 2026Only updates dependencies:@ai-sdk/provider 4.0.25@ai-sdk/provider-utils 5.0.57@ai-sdk/openai 4.0.89@ai-sdk/anthropic 4.0.76
5.0.110
7 Oct 2026amazon-bedrock,anthropic-aws: Skip SigV4 signing for non-ASCII header values
Patch Changes
amazon-bedrock,anthropic-aws: Skip SigV4 signing for non-ASCII header values
5.0.109
7 Oct 2026anthropic: Add Claude Haiku 5.5 support
Patch Changes
anthropic: Add Claude Haiku 5.5 support
add the
claude-haiku-5-5model ID to@ai-sdk/anthropicand@ai-sdk/google-vertex,anthropic.claude-haiku-5-5andus.anthropic.claude-haiku-5-5to@ai-sdk/amazon-bedrock, andanthropic/claude-haiku-5.5to@ai-sdk/gatewayrecognize
claude-haiku-5-5as a known model with a 128k max output token limit, adaptive thinking, and all five effort levels; thinking can be disabled up tohigheffort, andxhighandmaxare lowered tohighwith a warning when thinking is disabledclaude-haiku-5-5does not support thinking token budgets:thinking: { type: 'enabled', budgetTokens }is converted to adaptive thinking with a warninguse the JSON tool fallback for structured output on Amazon Bedrock for
claude-haiku-5-5
5.0.108
7 Oct 2026Only updates dependencies:@ai-sdk/provider 4.0.24@ai-sdk/openai 4.0.86@ai-sdk/anthropic 4.0.74@ai-sdk/provider-utils 5.0.56
5.0.107
6 Oct 2026provider: 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
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.
5.0.106
5 Oct 2026Only updates dependencies:@ai-sdk/provider 4.0.22@ai-sdk/provider-utils 5.0.54@ai-sdk/openai 4.0.84@ai-sdk/anthropic 4.0.72
5.0.105
1 Oct 2026Add
requestMetadatasupport to the Converse and ConverseStream APIs viaproviderOptions.amazonBedrock.requestMetadata. Accepts aRecord<string, string>for per-request tagging in Amazon Bedrock invocation logs for cost attribution.
5.0.104
30 Sep 2026Only updates a dependency:@ai-sdk/anthropic 4.0.71
5.0.103
30 Sep 2026Validate hostname parts using
isValidHostnamePart
Patch Changes
Validate hostname parts using
isValidHostnamePart
5.0.101 – 5.0.1022 releases that only update dependencies@ai-sdk/openai 4.0.82@ai-sdk/anthropic 4.0.69@ai-sdk/provider 4.0.20and 1 moreShow each
- 5.0.10230 Sep 2026@ai-sdk/openai 4.0.82@ai-sdk/anthropic 4.0.69@ai-sdk/provider 4.0.20and 1 more
- 5.0.10129 Sep 2026@ai-sdk/openai 4.0.81
5.0.100
28 Sep 2026Use standards-compliant User-Agent header
Patch Changes
Use standards-compliant User-Agent header
5.0.99
28 Sep 2026anthropic: Add Claude Sonnet 5.5 support
Patch Changes
anthropic: Add Claude Sonnet 5.5 support
add the
claude-sonnet-5-5model ID to@ai-sdk/anthropicand@ai-sdk/google-vertex,anthropic.claude-sonnet-5-5andus.anthropic.claude-sonnet-5-5to@ai-sdk/amazon-bedrock, andanthropic/claude-sonnet-5.5to@ai-sdk/gatewayadd the
between_toolsthinking type (thinking: { type: 'between_tools' }), the lowest thinking setting onclaude-sonnet-5-5;xhighandmaxeffort are lowered tohighwith a warning because the API rejects them withbetween_toolsclaude-sonnet-5-5rejects disabled thinking:thinking: { type: 'disabled' }is replaced withbetween_toolsthinking (with a warning),reasoning: 'none'maps tobetween_toolsthinking, and budget-based thinking is converted to adaptive thinkingclaude-sonnet-5-5rejects forced tool use:requiredand named tool choices fall back toauto, andstructuredOutputMode: 'jsonTool'falls back to native structured outputs, each with a warning
5.0.98
28 Sep 2026amazon-bedrock: Omit max output tokens for Nova 2 Lite high reasoning
provider: Advertise image model file and mask input support
Use confirmed model IDs for capability declarations so unrecognized model names remain unknown. Include Together AI FLUX.2 Pro and Flex single-image editing, and allow asynchronous capability lookups and middleware overrides to resolve to unknown.
Advertise QuiverAI Arrow 2 and Arrow 2 Telos file-input support, and mark Together AI Gemini image inputs unsupported by the current single-image request mapping.
Patch Changes
amazon-bedrock: Omit max output tokens for Nova 2 Lite high reasoning
provider: Advertise image model file and mask input support
Use confirmed model IDs for capability declarations so unrecognized model names remain unknown. Include Together AI FLUX.2 Pro and Flex single-image editing, and allow asynchronous capability lookups and middleware overrides to resolve to unknown.
Advertise QuiverAI Arrow 2 and Arrow 2 Telos file-input support, and mark Together AI Gemini image inputs unsupported by the current single-image request mapping.
5.0.94 – 5.0.973 releases that only update dependencies@ai-sdk/openai 4.0.78@ai-sdk/provider-utils 5.0.49@ai-sdk/anthropic 4.0.65Show each
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.