Skip to content
Permalink

Comparing changes

Choose two branches to see what’s changed or to start a new pull request. If you need to, you can also or learn more about diff comparisons.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also . Learn more about diff comparisons here.
base repository: vercel/ai
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: @ai-sdk/mcp@1.0.85
Choose a base ref
...
head repository: vercel/ai
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: @ai-sdk/mcp@1.0.86
Choose a head ref
  • 7 commits
  • 80 files changed
  • 6 contributors

Commits on Sep 24, 2026

  1. Backport: feat(openai): support message-level reasoning effort updates (

    #21421)
    
    Backports #21418 to `release-v6.0`, adding positioned reasoning effort
    updates on empty system messages for OpenAI and Azure Responses models.
    Includes Azure namespace fallback and type exports, examples,
    documentation, and patch changesets.
    
    The implementation uses the v3 provider interface, v6 Responses option
    schemas, and `providerOptions.openai.reasoningEffort`. Examples use
    `response.messages`. Existing system-message defaults and request-level
    warning behavior are preserved. Compaction settings that do not exist in
    v6 are omitted. Core normalization coverage uses the existing mock model
    without adding a provider dependency to `ai`.
    
    Validation:
    - OpenAI Responses suites: 501 tests passed in each of Node and Edge,
    including 76 positioned-update cases.
    - Azure provider: 72 tests passed in each runtime.
    - Core normalization: 2 tests passed in each runtime.
    - `pnpm type-check:full` passed, including examples.
    - `pnpm check` and `git diff --check` passed.
    
    ---------
    
    Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
    Co-authored-by: Gregor Martynus <39992+gr2m@users.noreply.github.com>
    github-actions[bot] and gr2m authored Sep 24, 2026
    Configuration menu
    Copy the full SHA
    0fb3a22 View commit details
    Browse the repository at this point in the history
  2. Backport: chore(provider/gateway): update gateway model settings file…

    …s v6 (#21450)
    
    Automated update of the gateway model settings files.
    
    Regenerated from the upstream gateway model list by
    [`update-model-settings.yml`](https://github.com/vercel/ai/blob/main/.github/workflows/update-model-settings.yml)
    ([this run](https://github.com/vercel/ai/actions/runs/36026668330)).
    Merging this PR is what surfaces newly-released gateway models as IDE
    autocomplete suggestions when users type a model string with
    `@ai-sdk/gateway`.
    
    See the workflow file for details on cadence and scope.
    
    Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
    github-actions[bot] authored Sep 24, 2026
    Configuration menu
    Copy the full SHA
    1915a6f View commit details
    Browse the repository at this point in the history
  3. Version Packages (#21423)

    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 release-v6.0, this
    PR will be updated.
    
    
    # Releases
    ## ai@6.0.291
    
    ### Patch Changes
    
    - Updated dependencies [1915a6f]
      - @ai-sdk/gateway@3.0.201
    ## @ai-sdk/amazon-bedrock@4.0.183
    
    ### Patch Changes
    
    - Updated dependencies [0fb3a22]
      - @ai-sdk/openai@3.0.118
    ## @ai-sdk/angular@2.0.292
    
    ### Patch Changes
    
    - ai@6.0.291
    ## @ai-sdk/azure@3.0.126
    
    ### Patch Changes
    
    - 0fb3a22: Export `OpenAIResponsesSystemMessageOptions` for typed
    message-level reasoning effort updates in Azure Responses models.
    - Updated dependencies [0fb3a22]
      - @ai-sdk/openai@3.0.118
    ## @ai-sdk/gateway@3.0.201
    
    ### Patch Changes
    
    - 1915a6f: Backport: chore(provider/gateway): update gateway model
    settings files
    ## @ai-sdk/langchain@2.0.299
    
    ### Patch Changes
    
    - ai@6.0.291
    ## @ai-sdk/llamaindex@2.0.291
    
    ### Patch Changes
    
    - ai@6.0.291
    ## @ai-sdk/openai@3.0.118
    
    ### Patch Changes
    
    - 0fb3a22: The OpenAI Responses provider now accepts
    `providerOptions.openai.reasoningEffortUpdate` on empty system messages
    and sends each update at its position in the conversation. This lets
    applications change reasoning effort during a conversation while
    preserving the prompt prefix for caching.
    ## @ai-sdk/react@3.0.294
    
    ### Patch Changes
    
    - ai@6.0.291
    ## @ai-sdk/rsc@2.0.291
    
    ### Patch Changes
    
    - ai@6.0.291
    ## @ai-sdk/svelte@4.0.291
    
    ### Patch Changes
    
    - ai@6.0.291
    ## @ai-sdk/vue@3.0.291
    
    ### Patch Changes
    
    - ai@6.0.291
    
    Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
    github-actions[bot] authored Sep 24, 2026
    Configuration menu
    Copy the full SHA
    9ddbf0e View commit details
    Browse the repository at this point in the history
  4. [v6.0] fix: preserve explicit MCP stdio environment values over inher…

    …ited parent defaults (#21465)
    
    ## Background
    
    Explicit MCP stdio environment values such as PATH, HOME, TEMP, and
    USERPROFILE were replaced by parent-process values, preventing sandboxed
    homes, private temp directories, and pinned executable paths.
    
    ## Root Cause
    
    getEnvironment copied custom values before unconditionally assigning
    inherited defaults from process.env; the original public-API
    reproduction confirmed those later assignments reached the child instead
    of explicit values.
    
    ## Summary
    
    Inherited defaults are now skipped when a matching custom key exists,
    including case-insensitive key matching on Windows, while unrelated
    defaults remain inherited. Added a patch changeset for @ai-sdk/mcp.
    
    ## Testing
    
    Added regression tests for explicit inherited-key overrides and Windows
    Path casing while retaining the existing immutability coverage.
    
    ## End-to-end Validation
    
    - `replay_original_reproduction`: passed with `no-longer-reproduces`;
    the MCP child received every explicit environment value.
    
    ## Related Issues
    
    Fixes #21430
    
    Closes #21462
    
    Backport of #21435
    
    Co-authored-by: ai-sdk-factory[bot] <305873210+ai-sdk-factory[bot]@users.noreply.github.com>
    Co-authored-by: ai-sdk-factory <308175966+ai-sdk-factory@users.noreply.github.com>
    Co-authored-by: ohad6k <190882507+ohad6k@users.noreply.github.com>
    3 people authored Sep 24, 2026
    Configuration menu
    Copy the full SHA
    86fcba3 View commit details
    Browse the repository at this point in the history
  5. Configuration menu
    Copy the full SHA
    af4a3e4 View commit details
    Browse the repository at this point in the history

Commits on Sep 25, 2026

  1. Backport: Browserbase Search and Fetch gateway tools to v6 (#21495)

    ## Background
    
    Backports #20248 to `release-v6.0`, adding
    `gateway.tools.browserbaseSearch()` and
    `gateway.tools.browserbaseFetch()` for AI SDK v6.
    
    ## Summary
    
    - Include the Browserbase tool definitions, input/output schemas,
    documentation, generate/stream examples, tests, and patch changeset from
    the original PR.
    - Use v6's `createProviderToolFactoryWithOutputSchema` and existing Zod
    import pattern; update factory assertions to match v6's tool shape.
    - Use `result.fullStream` in the streaming examples.
    
    ## Validation
    
    - `pnpm --filter '@ai-sdk/gateway...' --filter 'ai...' build`
    - `pnpm --filter @ai-sdk/gateway test` — 439 tests passed in Node and
    439 in Edge.
    - `pnpm type-check:full`
    - `pnpm check`
    - `git diff --check`
    
    ## Manual Verification
    
    Built the v6 `ai` and `@ai-sdk/gateway` packages and exercised their
    public tool exports with `generateText` and `streamText`. All eight
    cases passed: search plus raw, Markdown, and schema-based JSON fetch in
    each mode. Confirmed configuration overrides, output schema validation,
    and `validateUIMessages` compatibility.
    
    ## Checklist
    
    - [x] Tests have been added / updated (for bug fixes / features)
    - [x] Documentation has been added / updated (for bug fixes / features)
    - [x] A _patch_ changeset for relevant packages has been added (for bug
    fixes / features - run `pnpm changeset` in the project root)
    - [x] I have reviewed this pull request (self-review)
    ZYPX authored Sep 25, 2026
    Configuration menu
    Copy the full SHA
    2cf7cb4 View commit details
    Browse the repository at this point in the history
  2. Version Packages (#21468)

    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 release-v6.0, this
    PR will be updated.
    
    
    # Releases
    ## ai@6.0.292
    
    ### Patch Changes
    
    - Updated dependencies [2cf7cb4]
    - Updated dependencies [af4a3e4]
      - @ai-sdk/gateway@3.0.202
    ## @ai-sdk/angular@2.0.293
    
    ### Patch Changes
    
    - ai@6.0.292
    ## @ai-sdk/gateway@3.0.202
    
    ### Patch Changes
    
    - 2cf7cb4: feat (provider/gateway): add Browserbase Search and Fetch
    tool support
    - af4a3e4: feat(provider/gateway): accept structured-output in the has
    provider option
    ## @ai-sdk/langchain@2.0.300
    
    ### Patch Changes
    
    - ai@6.0.292
    ## @ai-sdk/llamaindex@2.0.292
    
    ### Patch Changes
    
    - ai@6.0.292
    ## @ai-sdk/mcp@1.0.86
    
    ### Patch Changes
    
    - 86fcba3: fix(mcp): preserve explicit stdio transport environment
    values
    ## @ai-sdk/react@3.0.295
    
    ### Patch Changes
    
    - ai@6.0.292
    ## @ai-sdk/rsc@2.0.292
    
    ### Patch Changes
    
    - ai@6.0.292
    ## @ai-sdk/svelte@4.0.292
    
    ### Patch Changes
    
    - ai@6.0.292
    ## @ai-sdk/vue@3.0.292
    
    ### Patch Changes
    
    - ai@6.0.292
    
    Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
    github-actions[bot] authored Sep 25, 2026
    Configuration menu
    Copy the full SHA
    0f3bd70 View commit details
    Browse the repository at this point in the history
Loading