NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
npm · #2684 most downloaded on npm
AI SDK by Vercel - build apps like ChatGPT, Claude, Gemini, and more with a single interface for any model using the Vercel AI Gateway or go direct to OpenAI, Anthropic, Google, or any other model provider.
Last release today
06 Oct 2026
Ships on a steady schedule
a new release about every 8 days
Some releases are documented
notes for 27 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
13 years old
1572 releases · first in 2014
89b7552b: chore (ai): remove deprecation from ai/react imports, add experimental_useObject
0565cd72: feat (ai/core): add toJsonResponse to generateObject result.
One column per quarter.
008725ec: feat (ai): add textStream, toTextStreamResponse(), and pipeTextStreamToResponse() to streamObject
85ef6d18: chore (ai): AI SDK 3.2 release
85712895: chore (@ai-sdk/provider-utils): move test helper to provider utils
4728c37f: feat (core): add text embedding model support to provider registry
06123501: feat (core): support https and data url strings in image parts
d25566ac: feat (core): add cosineSimilarity helper function
6fb14b5d: chore (streams): deprecate nanoid export.
3cabf078: fix(ai/rsc): Refactor streamable UI internal implementation
85f209a4: chore: extracted ui library support into separate modules
fcf4323b: fix (core): filter out empty assistant text messages
28427d3e: feat (core): add streamObject onFinish callback
102ca22f: feat (core): add object promise to streamObject result
c9198d4d: feat (ui): send annotation and data fields in useChat when sendExtraMessageFields is true
5ee44cae: feat (provider): langchain StringOutputParser support
ff281126: fix(ai/rsc): Remove extra reconcilation of streamUI
93cae126: fix(ai/rsc): Fix unsafe {} type in application code for StreamableValue
c03cafe6: chore (core, ui): rename maxAutomaticRoundtrips to maxToolRoundtrips
14bb8694: chore (ui): move maxAutomaticRoundtrips and addToolResult out of experimental
09698bca: chore (streams): deprecate streaming helpers that have a provider replacement
0e1da476: feat (core): add maxAutomaticRoundtrips setting to generateText
9882d24b: fix (ui/svelte): send data to server
f9dee8ac: fix(ai/rsc): Fix types for createStreamableValue and createStreamableUI
7de628e9: chore (ui): deprecate old function/tool call handling
f39c0dd2: feat (core, rsc): add toolChoice setting
8e780288: feat (ai/core): add onFinish callback to streamText
6109c6a: feat (ai/react): add experimental_maxAutomaticRoundtrips to useChat
60117c9: dependencies (ai/ui): add React 18.3 and 19 support (peer dependency)
ae05fb7: feat (ai/streams): add StreamData support to streamToResponse
a085d42: fix (ai/ui): decouple StreamData chunks from LLM stream
3a21030: feat (ai/core): add embedMany function
18a9655: feat (ai/svelte): add useAssistant
0f6bc4e: feat (ai/core): add embed function
f617b97: feat (ai): support client/server tool calls with useChat and streamText
2e78acb: Deprecate StreamingReactResponse (use AI SDK RSC instead).
5b01c13: feat (ai/core): add system message support in messages list
ceb44bc: feat (ai/ui): add stop() helper to useAssistant (important: AssistantResponse now requires OpenAI SDK 4.42+)
970a099: fix (ai/core): streamObject fixes partial json with empty objects correctly
d1b1880: fix (ai/core): allow reading streams in streamText result multiple times
0f77132: ai/rsc: remove experimental\_ from streamUI
00fe856 : feat(deepgram): transcription option fixes + speech voice/language composition, usage metadata, speed passthrough, and error parsing
00fe856: feat(deepgram): transcription option fixes + speech voice/language composition, usage metadata, speed passthrough, and error parsing
Transcription:
keyterm, paragraphs, intents, sentiment, and replace wereproviderOptions.deepgram but silently dropped from the/v1/listen request. They are now sent as query parameters. Also widens'nova-3' to any transcriptiondiarize no longer defaults to true. Speakerdiarize=true on every pre-recorded request unless explicitly optedproviderOptions.deepgram. Users who relied on the old default mustproviderOptions: { deepgram: { diarize: true } }.Speech:
aura-2, aura) compose the upstream model IDgenerateSpeech voice and language options<family>-<voice>-<language>, language defaults to en) and requirevoice; full voice IDs (e.g. aura-2-helena-en) keep passing throughDeepgramSpeechModelId union is trimmed to the familyproviderMetadata.deepgram carries modelName, modelUuid,additionalModelUuids, charCount (the billed character count),breaksApplied, pronunciationsApplied, pronunciationWarnings (whenrequestId from the /v1/speak response headers.speed option is passed through to Deepgram's speed parameter{ "err_code", "err_msg", "request_id" }APICallError.message carries the real cause instead of{ "error": { "message", "code" } }Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
41d5736: ai/core: re-expose language model types.
b9a831e: ai/rsc: add experimental_streamUI()
Updated dependencies [ 90192f1 ]
0e0d2af: ai/core: add pipeTextStreamToResponse helper to streamText.
74c63b1: ai/core: add toAIStreamResponse() helper to streamText.
e7e5898: use-assistant: fix missing message content
Updated dependencies [ b74971f ]
d6431ae: ai/core: add logprobs support (thanks @SamStenner for the contribution)
Updated dependencies [ d3bcad9 ]
f90f6a1: ai/core: add pipeAIStreamToResponse() to streamText result.
Updated dependencies [ 2959d35 ]
Updated dependencies [ 04be48f ]
onSetAIState and onGetUIState stableUpdated dependencies [ 8c65988 ]
Your coding agent can read these notes before it upgrades. Set up the MCP server →