Version Packages (#19362)

This PR was opened by the [Changesets
release](https://github.com/changesets/action) GitHub action. When
you're ready to do a release, you can merge this and the packages will
be published to npm automatically. If you're not ready to do a release
yet, that's fine, whenever you add more changesets to main, this PR will
be updated.


# Releases
## @ai-sdk/harness-cursor@1.0.0

### Major Changes

- ac1384b: feat(harness-cursor): implement Cursor harness adapter

### Patch Changes

- Updated dependencies [79186d1]
- Updated dependencies [b6396ff]
- Updated dependencies [8a15038]
  - @ai-sdk/harness-acp@1.0.25
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-fx@1.0.0

### Major Changes

- d0b241d: feat(harness-fx): implement fx harness adapter

### Patch Changes

- Updated dependencies [79186d1]
- Updated dependencies [b6396ff]
- Updated dependencies [8a15038]
  - @ai-sdk/harness-acp@1.0.25
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## ai@7.0.79

### Patch Changes

- b251584: Preserve active text and reasoning parts when another merged
UI message stream finishes a step, and align workflow stream
normalization with the explicit part end chunks.
- 591d25b: feat: add batch completion webhooks.
`experimental_startTextBatch` accepts a `webhookUrl`, and the gateway
provider registers it through the batch `callbackUrl` contract and
exports typed async-job metadata. Direct Anthropic and OpenAI batch
providers return an unsupported warning when the option is provided.
- 9de0baf: Parse structured generateText output when providers omit
finishReason but return text.
- Updated dependencies [591d25b]
  - @ai-sdk/gateway@4.0.64
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/alibaba@2.0.35

### Patch Changes

- 3789213: feat(provider/alibaba): support wan3 all-in-one video
generation
- 2214258: Prevent negative text output token counts when providers
report reasoning tokens. Perplexity reasoning tokens are now treated as
separate from completion tokens.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/amazon-bedrock@5.0.62

### Patch Changes

- Updated dependencies [591d25b]
- Updated dependencies [2214258]
  - @ai-sdk/anthropic@4.0.42
  - @ai-sdk/openai@4.0.47
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/angular@3.0.79

### Patch Changes

- 4396fe0: Avoid deep-cloning chat messages before submission so
non-cloneable metadata does not prevent requests.
- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/anthropic@4.0.42

### Patch Changes

- 591d25b: feat: add batch completion webhooks.
`experimental_startTextBatch` accepts a `webhookUrl`, and the gateway
provider registers it through the batch `callbackUrl` contract and
exports typed async-job metadata. Direct Anthropic and OpenAI batch
providers return an unsupported warning when the option is provided.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/anthropic-aws@2.0.34

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/anthropic@4.0.42
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/assemblyai@3.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/azure@4.0.49

### Patch Changes

- 0e36ab5: Warn and omit deprecated or ineffective DeepSeek sampling
options while
  preserving penalty sampling for Azure-hosted DeepSeek models.
- Updated dependencies [0e36ab5]
- Updated dependencies [068be49]
- Updated dependencies [56d47f5]
- Updated dependencies [49ba214]
- Updated dependencies [1ca2519]
- Updated dependencies [591d25b]
- Updated dependencies [f2a2598]
- Updated dependencies [f8e24b6]
- Updated dependencies [320b700]
- Updated dependencies [f70bd8a]
- Updated dependencies [2214258]
- Updated dependencies [f1cadd0]
- Updated dependencies [01fa8de]
- Updated dependencies [348dacb]
- Updated dependencies [560fc39]
- Updated dependencies [2814b2d]
  - @ai-sdk/deepseek@3.0.32
  - @ai-sdk/openai@4.0.47
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/baseten@2.1.15

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/black-forest-labs@2.0.31

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/bytedance@2.0.33

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/cartesia@3.0.25

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/cerebras@3.0.37

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/code-mode@1.0.36

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
## @ai-sdk/codemod@4.0.1

### Patch Changes

- e493ed7: Fix codemod scaffolding when target directories are missing.
## @ai-sdk/cohere@4.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/deepgram@3.1.1

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/deepinfra@3.0.37

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/deepseek@3.0.32

### Patch Changes

- 0e36ab5: Warn and omit deprecated or ineffective DeepSeek sampling
options while
  preserving penalty sampling for Azure-hosted DeepSeek models.
- 068be49: feat(provider/deepseek): add the `userId` chat option and
serialize it as `user_id`
- 56d47f5: feat(provider/deepseek): add DeepSeek V4 Flash and Pro model
IDs
- 49ba214: Validate DeepSeek file upload formats, sizes, and filename
lengths before sending requests.
- 1ca2519: Add provider-specific names for DeepSeek system, user, and
assistant messages.
- f2a2598: Validate DeepSeek strict tool calls against beta endpoint
requirements.
- f8e24b6: Preserve documented DeepSeek chat response metadata for
generate and stream.
- 320b700: Add DeepSeek chat completion log probability request and
response support.
- f70bd8a: Add support for DeepSeek beta assistant prefix completion
through message provider options.
- 2214258: Prevent negative text output token counts when providers
report reasoning tokens. Perplexity reasoning tokens are now treated as
separate from completion tokens.
- f1cadd0: Preserve the file object discriminator in DeepSeek upload
provider metadata.
- 01fa8de: Validate DeepSeek file response discriminators and numeric
metadata while preserving compatibility with omitted optional metadata.
- 348dacb: Expose DeepSeek response system fingerprints in provider
metadata for generated and streamed text.
- 560fc39: Align DeepSeek thinking and reasoning effort options with the
documented API values while mapping legacy options with compatibility
warnings.
- 2814b2d: Add DeepSeek image detail and inline file-data content part
support.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/devtools@1.0.13

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
## @ai-sdk/elevenlabs@3.0.31

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/fal@3.0.31

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/fireworks@3.0.40

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/fish-audio@3.0.8

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/gateway@4.0.64

### Patch Changes

- 591d25b: feat: add batch completion webhooks.
`experimental_startTextBatch` accepts a `webhookUrl`, and the gateway
provider registers it through the batch `callbackUrl` contract and
exports typed async-job metadata. Direct Anthropic and OpenAI batch
providers return an unsupported warning when the option is provided.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/gladia@3.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/gmicloud@3.0.8

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/google@4.0.51

### Patch Changes

- e7fc90e: feat(google): support the Gemini Batch API with
experimental_startTextBatch
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/google-vertex@5.0.63

### Patch Changes

- Updated dependencies [e7fc90e]
- Updated dependencies [591d25b]
  - @ai-sdk/google@4.0.51
  - @ai-sdk/anthropic@4.0.42
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/groq@4.0.31

### Patch Changes

- 2214258: Prevent negative text output token counts when providers
report reasoning tokens. Perplexity reasoning tokens are now treated as
separate from completion tokens.
- 9e9bf11: fix: disable reasoning for supported Groq models when
top-level reasoning is set to `none`
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness@1.0.87

### Patch Changes

- 8a15038: feat(harness): add `credentialForwarding` setting to bridge
backed harness adapters for granular control
- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-acp@1.0.25

### Patch Changes

- 79186d1: Allow pinned OpenCode and Grok Build install scripts during
pnpm 11 sandbox bootstrap.
- b6396ff: feat(harness-acp): support `title` and `toolUseKind` for ACP
tools
- 8a15038: feat(harness): add `credentialForwarding` setting to bridge
backed harness adapters for granular control
- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-claude-code@1.0.90

### Patch Changes

- 8a15038: feat(harness): add `credentialForwarding` setting to bridge
backed harness adapters for granular control
- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-cline@1.0.14

### Patch Changes

- Updated dependencies [591d25b]
- Updated dependencies [8a15038]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-codex@1.0.89

### Patch Changes

- 8a15038: feat(harness): add `credentialForwarding` setting to bridge
backed harness adapters for granular control
- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-deepagents@1.0.87

### Patch Changes

- 8a15038: feat(harness): add `credentialForwarding` setting to bridge
backed harness adapters for granular control
- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-grok-build@1.0.24

### Patch Changes

- 79186d1: Allow pinned OpenCode and Grok Build install scripts during
pnpm 11 sandbox bootstrap.
- 8a15038: feat(harness): add `credentialForwarding` setting to bridge
backed harness adapters for granular control
- Updated dependencies [79186d1]
- Updated dependencies [b6396ff]
- Updated dependencies [8a15038]
  - @ai-sdk/harness-acp@1.0.25
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-opencode@1.0.88

### Patch Changes

- 79186d1: Allow pinned OpenCode and Grok Build install scripts during
pnpm 11 sandbox bootstrap.
- 8a15038: feat(harness): add `credentialForwarding` setting to bridge
backed harness adapters for granular control
- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/harness-pi@1.0.89

### Patch Changes

- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/huggingface@2.0.37

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/hume@3.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/klingai@4.0.32

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/langchain@3.0.79

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
## @ai-sdk/llamaindex@3.0.79

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
## @ai-sdk/lmnt@3.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/luma@3.0.31

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/mcp@2.0.37

### Patch Changes

- 1175434: Fetch all paginated tool definitions when creating an MCP
tool set.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/minimax@3.0.18

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/anthropic@4.0.42
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/mistral@4.0.33

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/moonshotai@3.0.38

### Patch Changes

- 2214258: Prevent negative text output token counts when providers
report reasoning tokens. Perplexity reasoning tokens are now treated as
separate from completion tokens.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/open-responses@2.0.31

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/openai@4.0.47

### Patch Changes

- 591d25b: feat: add batch completion webhooks.
`experimental_startTextBatch` accepts a `webhookUrl`, and the gateway
provider registers it through the batch `callbackUrl` contract and
exports typed async-job metadata. Direct Anthropic and OpenAI batch
providers return an unsupported warning when the option is provided.
- 2214258: Prevent negative text output token counts when providers
report reasoning tokens. Perplexity reasoning tokens are now treated as
separate from completion tokens.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/openai-compatible@3.0.37

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/otel@1.0.79

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider@4.0.8
## @ai-sdk/perplexity@4.0.32

### Patch Changes

- 2214258: Prevent negative text output token counts when providers
report reasoning tokens. Perplexity reasoning tokens are now treated as
separate from completion tokens.
- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/policy-opa@1.0.79

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/prodia@2.0.31

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/provider@4.0.8

### Patch Changes

- 591d25b: feat: add batch completion webhooks.
`experimental_startTextBatch` accepts a `webhookUrl`, and the gateway
provider registers it through the batch `callbackUrl` contract and
exports typed async-job metadata. Direct Anthropic and OpenAI batch
providers return an unsupported warning when the option is provided.
## @ai-sdk/provider-utils@5.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
## @ai-sdk/quiverai@2.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/react@4.0.82

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [1175434]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/mcp@2.0.37
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/replicate@3.0.31

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/revai@3.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/rsc@3.0.79

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/sandbox-just-bash@1.0.87

### Patch Changes

- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/sandbox-vercel@1.0.87

### Patch Changes

- 7ab0f88: Add support for `@vercel/sandbox` v3 while preserving the
adapter's existing Node 24 runtime default.
- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/svelte@5.0.79

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/togetherai@3.0.38

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/openai-compatible@3.0.37
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/tui@1.0.80

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
## @ai-sdk/valibot@3.0.30

### Patch Changes

- @ai-sdk/provider-utils@5.0.30
## @ai-sdk/voyage@2.0.30

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/vue@4.0.79

### Patch Changes

- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/workflow@2.0.9

### Patch Changes

- b251584: Preserve active text and reasoning parts when another merged
UI message stream finishes a step, and align workflow stream
normalization with the explicit part end chunks.
- Updated dependencies [b251584]
- Updated dependencies [591d25b]
- Updated dependencies [9de0baf]
  - ai@7.0.79
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30
## @ai-sdk/workflow-harness@1.0.87

### Patch Changes

- Updated dependencies [8a15038]
  - @ai-sdk/harness@1.0.87
## @ai-sdk/xai@4.0.44

### Patch Changes

- Updated dependencies [591d25b]
  - @ai-sdk/provider@4.0.8
  - @ai-sdk/provider-utils@5.0.30

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>