#21256

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-sdk/alibaba@1.0.57

### Patch Changes

- 3e59b31: fix(alibaba): use model-specific structured output modes

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

…21287)

## Background

Backport of #21286 to `release-v6.0`.


Claude Opus 5.5 (`claude-opus-5-5`) rejects several request shapes that
`@ai-sdk/anthropic` previously sent unconditionally: `thinking: { type:
'disabled' }` and budget-based thinking, forced `tool_choice`
(`required` or a named tool), and the older `computer_*` tool types.
Without these changes, common SDK usage 400s against the new model.

## Summary

- Add the `claude-opus-5-5` model ID to `@ai-sdk/anthropic` and
`anthropic/claude-opus-5.5` to `@ai-sdk/gateway`.
- Models that always use adaptive thinking (`claude-opus-5-5`,
`claude-fable-5`, `claude-fable-5-1`) no longer receive disabled or
budget-based thinking. The provider drops the unsupported setting, maps
`reasoning: 'none'` to `effort: 'low'`, and emits a warning.
- Models that reject forced tool use (`claude-opus-5-5`,
`claude-fable-5-1`) fall back to `auto` tool choice for `required` and
named tool choices, and to native structured outputs when
`structuredOutputMode: 'jsonTool'` is requested, each with a warning.
- Add the `computerToolset_20260801` computer use tool
(`computer_toolset_20260801`), required for computer use on
`claude-opus-5-5`. Member calls are mapped to `action` and
`toolset_name` round-trips through tool results.
- Use the documented `mid-conversation-output-config-2026-07-01` beta
header for per-message effort.
- Docs and examples for the new model and tool.

## End-to-End Verification

Ran the new examples under `examples/ai-functions/src` against the
Anthropic API with `claude-opus-5-5`:

- `generate-text/anthropic/claude-opus-5-5.ts` (default adaptive
thinking)
- `generate-text/anthropic/claude-opus-5-5-low-effort.ts` (`reasoning:
'none'` mapped to low effort with warning)
- `stream-text/anthropic/claude-opus-5-5-thinking-updates.ts`
(per-message effort beta)
- `stream-text/anthropic/computer-toolset.ts`
(`computer_toolset_20260801` round-trip)

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

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@6.0.288

### Patch Changes

- Updated dependencies [bd8bf1f]
  - @ai-sdk/gateway@3.0.198
## @ai-sdk/amazon-bedrock@4.0.180

### Patch Changes

- Updated dependencies [bd8bf1f]
  - @ai-sdk/anthropic@3.0.120
## @ai-sdk/angular@2.0.289

### Patch Changes

- ai@6.0.288
## @ai-sdk/anthropic@3.0.120

### Patch Changes

- bd8bf1f: feat(anthropic): add Claude Opus 5.5 support

- add the `claude-opus-5-5` model ID to `@ai-sdk/anthropic` and
`anthropic/claude-opus-5.5` to `@ai-sdk/gateway`
- models that always use adaptive thinking (`claude-opus-5-5`,
`claude-fable-5`, `claude-fable-5-1`) no longer receive `thinking: {
type: 'disabled' }` or budget-based thinking; the provider drops the
unsupported setting and emits a warning
- models that reject forced tool use (`claude-opus-5-5`,
`claude-fable-5-1`) fall back to `auto` tool choice for `required` and
named tool choices, and to native structured outputs when
`structuredOutputMode: 'jsonTool'` is requested, each with a warning
- add the `computerToolset_20260801` computer use tool
(`computer_toolset_20260801`), which is required for computer use on
`claude-opus-5-5`
- use the documented `mid-conversation-output-config-2026-07-01` beta
header for per-message effort
## @ai-sdk/anthropic-aws@1.0.42

### Patch Changes

- Updated dependencies [bd8bf1f]
  - @ai-sdk/anthropic@3.0.120
## @ai-sdk/gateway@3.0.198

### Patch Changes

- bd8bf1f: feat(anthropic): add Claude Opus 5.5 support

- add the `claude-opus-5-5` model ID to `@ai-sdk/anthropic` and
`anthropic/claude-opus-5.5` to `@ai-sdk/gateway`
- models that always use adaptive thinking (`claude-opus-5-5`,
`claude-fable-5`, `claude-fable-5-1`) no longer receive `thinking: {
type: 'disabled' }` or budget-based thinking; the provider drops the
unsupported setting and emits a warning
- models that reject forced tool use (`claude-opus-5-5`,
`claude-fable-5-1`) fall back to `auto` tool choice for `required` and
named tool choices, and to native structured outputs when
`structuredOutputMode: 'jsonTool'` is requested, each with a warning
- add the `computerToolset_20260801` computer use tool
(`computer_toolset_20260801`), which is required for computer use on
`claude-opus-5-5`
- use the documented `mid-conversation-output-config-2026-07-01` beta
header for per-message effort
## @ai-sdk/google-vertex@4.0.203

### Patch Changes

- Updated dependencies [bd8bf1f]
  - @ai-sdk/anthropic@3.0.120
## @ai-sdk/langchain@2.0.296

### Patch Changes

- ai@6.0.288
## @ai-sdk/llamaindex@2.0.288

### Patch Changes

- ai@6.0.288
## @ai-sdk/minimax@2.0.18

### Patch Changes

- Updated dependencies [bd8bf1f]
  - @ai-sdk/anthropic@3.0.120
## @ai-sdk/react@3.0.291

### Patch Changes

- ai@6.0.288
## @ai-sdk/rsc@2.0.288

### Patch Changes

- ai@6.0.288
## @ai-sdk/svelte@4.0.288

### Patch Changes

- ai@6.0.288
## @ai-sdk/vue@3.0.288

### Patch Changes

- ai@6.0.288

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

Backport of the primary AI SDK 7 PR:
#21305.

Adds GPT-6 Sol and Luna model IDs to OpenAI and Gateway, including
Gateway fast variants, and preserves `reasoningEffort: 'none'` for both
models. Includes regression tests, documentation, examples, and a patch
changeset.

Validation: OpenAI and Gateway builds and Node/Edge tests, full
workspace type checking, and targeted lint/format checks passed.

…nsupported tool use (#21312)

## Background

Amazon Bedrock Claude Opus 5.5 structured object generation returned
HTTP 400 because the SDK forced unsupported tool use.

## Root Cause

The Bedrock structured-output path ignored the model's
rejectsForcedToolUse capability and only allowed JSON-instruction
fallback when user tools existed, causing schema requests to generate a
synthetic required JSON tool with toolChoice.any.

## Summary

Bedrock now uses schema instructions for models that reject forced
tools, and Opus 5 and 5.5 model IDs were added to the Bedrock and Vertex
Anthropic types.

## Testing

Added regression coverage verifying that default and explicit jsonTool
modes omit forced tools and native output_config while returning
schema-shaped JSON.

## End-to-end Validation

- Immutable `pnpm -C examples/ai-functions exec tsx
src/reproduction/issue-21290-bedrock-opus-5-5-structured-output.ts`
replay exited 0 and returned the expected structured output without the
original failure signal.

## Related Issues

Fixes #21290

Closes #21311

Backport of #21297

---------

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: Gregor Martynus <39992+gr2m@users.noreply.github.com>

…ing guardrails (#21211)

## Background

The release-v6.0 middleware examples omitted the required V3
specification field, and the guardrails example did not safely handle
matches split across stream deltas.

## Summary

Added specificationVersion: 'v3' to all five middleware examples and
implemented per-text-block buffered streaming redaction with guidance
about latency, memory use, split matches, and unbounded patterns.

## Testing

pnpm check and pnpm type-check:full passed. A direct file-targeted
formatter invocation was unsupported, but the repository-wide formatting
and lint check passed.

## Related Issues

Fixes #21209

Co-authored-by: ai-sdk-factory <308175966+ai-sdk-factory@users.noreply.github.com>
Co-authored-by: roshcompanylabs <144812571+roshcompanylabs@users.noreply.github.com>

…c chat submission (#21327)

## Background

Preliminary static and dynamic tool outputs were treated as complete,
allowing sendAutomaticallyWhen to submit an unfinished chat.

## Root Cause

Both completion helpers accepted every output-available tool part
without checking its preliminary flag; the original reproduction
confirmed this caused premature transport requests.

## Summary

Updated both chat completion helpers to require non-preliminary tool
outputs and added a patch changeset.

## Testing

Added regression tests covering preliminary outputs from static and
dynamic tools for both tool-call and approval completion helpers.

## End-to-end Validation

- `replay_original_reproduction`: the immutable reproduction exited
successfully with no premature automatic submission
(`no-longer-reproduces`).

## Related Issues

Fixes #21083

Closes #21321

Backport of #21087

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: brennanbutler01 <64561607+brennanbutler01@users.noreply.github.com>

…21331)

## Background

Chat transports in the published ai package sent caller-configured
Content-Type headers as duplicate comma-separated values instead of a
single overridden value.

## Root Cause

normalizeHeaders lowercased configured header names, while
HttpChatTransport added its default as Content-Type. Both object
properties reached Fetch, which merged them case-insensitively; the
local HTTP reproduction and a focused Headers diagnostic confirmed the
combined value.

## Summary

Changed the default chat transport content-type key to lowercase so
normalized caller headers override it, and added an ai package patch
changeset.

## Testing

Added regression coverage for content-type overrides supplied through
constructor, per-request, and prepared-request headers.

## End-to-end Validation

- `pnpm -C examples/ai-functions exec tsx
src/reproduction/issue-21044-duplicate-content-type.ts` via immutable
replay exited successfully; all configured requests delivered one
Content-Type value and the original failure signal was absent.

## Related Issues

Fixes #21044

Closes #21313

Backport of #21054

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: brennanbutler01 <64561607+brennanbutler01@users.noreply.github.com>

## Problem

On Node.js 24.21.0, `streamText` can invoke `onInputAvailable` before
the final `onInputDelta` callbacks. Input availability runs in the
upstream tool transform while start/delta callbacks run downstream, so
their ordering depends on stream scheduling. This reproduces on the v6
branch and accounts for the callback snapshot failure in #21307.

## Change

Move start/delta callback invocation into the existing tool transform
and await them in stream order before input availability. Keep
downstream chunk formatting and the existing approval/execution
handling. This is a targeted fix rather than the full refactor from
#13681.

Add an async callback-order regression test and a patch changeset.

## Validation

- Confirmed the existing callback snapshot and new regression fail
without the fix on Node 24.21.0.
- Node 24.20.0: both affected test files pass, 403 tests.
- Node 24.21.0: callback tests pass in Node and Edge configurations;
both affected files have 398 passing tests and five pre-existing abort
snapshot failures.
- Full workspace type check and lint/format checks for changed files
pass.

The five abort snapshots are a separate scheduling-sensitive issue and
remain outside this callback-order fix. This PR does not resolve all
failures in the original CI run.

…oducing invalid requests (#21336)

## Background

Schema-less JSON output was unusable because Open Responses generated a
request format that the Responses API rejected before model execution.

## Root Cause

The provider unconditionally used the json_schema discriminator for JSON
output but conditionally omitted its required name and schema fields
when no schema was supplied; the reproduced request body and API error
confirmed this invalid branch.

## Summary

Schema-backed JSON retains json_schema, while schema-less JSON now uses
json_object. Internal request types include json_object support,
regression coverage was added, and a patch changeset documents the fix.

## Testing

Regression coverage verifies the schema-less json_object request shape
while existing tests preserve schema-backed behavior. All 107 Open
Responses tests passed in both Node and Edge, package type-check and
focused static checks passed, and the immutable reproduction no longer
reproduces the failure.

## End-to-end Validation

- `replay_original_reproduction` — exited successfully without the
original missing-name failure signal.
- `pnpm -C examples/ai-functions exec tsx -e ...` — the live OpenAI
Responses call returned `{"ok":true}` using `Output.json()`.

## Related Issues

Fixes #21202

Closes #21319

Backport of #21206

---------

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: Gregor Martynus <39992+gr2m@users.noreply.github.com>

Backports #21267. Preserves the original gs:// file URI string when URL
parsing would percent-encode literal characters, so Vertex serializes
the intended Cloud Storage object name. Tests: focused AI and Google
regression suites, package type checks for provider/ai/google, and pnpm
check.

…max_output_tokens (#21353)

## Background

When `max_output_tokens` is reached while the model is emitting a
function call, the Responses API closes the item with
`response.output_item.done` carrying `status: "incomplete"` and the
partial `arguments`, then ends the response with
`incomplete_details.reason: "max_output_tokens"`.

## Root Cause

The `function_call` (and `custom_tool_call`) item schema on
`response.output_item.done` required `status: z.literal('completed')`.
The incomplete item failed schema validation, the stream emitted a
`TypeValidationError` ("unhandled event") and ended with `finishReason:
'error'` instead of the expected `'length'`. Callers with a tool loop
saw an error rather than a truncated tool call. `main` already accepts
`incomplete` here (via #17790); this is the narrow schema portion
backported to the v6 line.

## Summary

- `openai-responses-api.ts`: `function_call` and `custom_tool_call`
items on `response.output_item.done` accept `status: 'in_progress' |
'completed' | 'incomplete'` (matching `response.output_item.added`).
- The incomplete call is emitted as a `tool-call` with the partial
arguments and the response finishes with `length`, the same shape other
providers produce when a tool call is truncated.
- Changeset: `@ai-sdk/openai` patch.

## Testing

New streaming test in `openai-responses-language-model.test.ts` with an
`incomplete` function_call + `response.incomplete { reason:
'max_output_tokens' }` fixture: asserts no `error` part, a `tool-call`
with the partial input, and `finishReason: { unified: 'length', raw:
'max_output_tokens' }`. Fails on the branch without the schema change
(stream yields an `error` part), passes with it. Full `@ai-sdk/openai`
node suite: 910 passed.

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@6.0.289

### Patch Changes

- 86f0089: fix(ai): prevent preliminary tool outputs from completing
chats
- da2e17b: fix(provider): preserve opaque file URI strings for provider
serialization
- f211ce5: fix(ai): prevent duplicate content types in chat transport
requests
- 2f1488e: Fix streamed tool input callbacks firing out of order on
newer Node.js versions by awaiting start and delta callbacks before
invoking `onInputAvailable`.
- Updated dependencies [da2e17b]
- Updated dependencies [0b9d00b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/gateway@3.0.199
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/alibaba@1.0.58

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/amazon-bedrock@4.0.181

### Patch Changes

- d0c9d86: fix(amazon-bedrock): avoid forced structured-output tools for
incompatible Claude models
- Updated dependencies [ad8de8d]
- Updated dependencies [da2e17b]
- Updated dependencies [0b9d00b]
  - @ai-sdk/openai@3.0.116
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/anthropic@3.0.121
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/angular@2.0.290

### Patch Changes

- Updated dependencies [86f0089]
- Updated dependencies [da2e17b]
- Updated dependencies [f211ce5]
- Updated dependencies [2f1488e]
  - ai@6.0.289
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/anthropic@3.0.121

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/anthropic-aws@1.0.43

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/anthropic@3.0.121
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/assemblyai@2.0.60

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/azure@3.0.124

### Patch Changes

- Updated dependencies [ad8de8d]
- Updated dependencies [da2e17b]
- Updated dependencies [0b9d00b]
  - @ai-sdk/openai@3.0.116
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/deepseek@2.0.66
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/baseten@1.1.14

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/black-forest-labs@1.0.63

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/bytedance@1.0.42

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/cerebras@2.0.83

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/cohere@3.0.63

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/deepgram@2.0.59

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/deepinfra@2.0.81

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/deepseek@2.0.66

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/devtools@0.0.27

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
## @ai-sdk/elevenlabs@2.0.59

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/fal@2.0.60

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/fireworks@2.0.87

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/fish-audio@2.0.12

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/gateway@3.0.199

### Patch Changes

- 0b9d00b: feat: GPT-6 Sol and Luna model IDs
- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/gladia@2.0.60

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/gmicloud@2.0.10

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/google@3.0.126

### Patch Changes

- da2e17b: fix(provider): preserve opaque file URI strings for provider
serialization
- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/google-vertex@4.0.204

### Patch Changes

- d0c9d86: fix(amazon-bedrock): avoid forced structured-output tools for
incompatible Claude models
- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/google@3.0.126
  - @ai-sdk/anthropic@3.0.121
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/groq@3.0.68

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/huggingface@1.0.79

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/hume@2.0.59

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/klingai@3.0.45

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/langchain@2.0.297

### Patch Changes

- Updated dependencies [86f0089]
- Updated dependencies [da2e17b]
- Updated dependencies [f211ce5]
- Updated dependencies [2f1488e]
  - ai@6.0.289
## @ai-sdk/llamaindex@2.0.289

### Patch Changes

- Updated dependencies [86f0089]
- Updated dependencies [da2e17b]
- Updated dependencies [f211ce5]
- Updated dependencies [2f1488e]
  - ai@6.0.289
## @ai-sdk/luma@2.0.59

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/mcp@1.0.84

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/minimax@2.0.19

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/anthropic@3.0.121
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/mistral@3.0.66

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/moonshotai@2.0.58

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/open-responses@1.0.44

### Patch Changes

- 7a1d302: fix(open-responses): use JSON object mode for schema-less
JSON output
- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/openai@3.0.116

### Patch Changes

- ad8de8d: fix(openai): accept `incomplete` function_call and
custom_tool_call items on the Responses stream, so a call truncated by
`max_output_tokens` finishes with `length` instead of a
`TypeValidationError`
- 0b9d00b: feat: GPT-6 Sol and Luna model IDs
- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/openai-compatible@2.0.77

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/perplexity@3.0.62

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/prodia@1.0.59

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/provider@3.0.17

### Patch Changes

- da2e17b: fix(provider): preserve opaque file URI strings for provider
serialization
## @ai-sdk/provider-utils@4.0.53

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
## @ai-sdk/quiverai@1.0.27

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/react@3.0.292

### Patch Changes

- Updated dependencies [86f0089]
- Updated dependencies [da2e17b]
- Updated dependencies [f211ce5]
- Updated dependencies [2f1488e]
  - ai@6.0.289
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/replicate@2.0.59

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/revai@2.0.59

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/rsc@2.0.289

### Patch Changes

- Updated dependencies [86f0089]
- Updated dependencies [da2e17b]
- Updated dependencies [f211ce5]
- Updated dependencies [2f1488e]
  - ai@6.0.289
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/svelte@4.0.289

### Patch Changes

- Updated dependencies [86f0089]
- Updated dependencies [da2e17b]
- Updated dependencies [f211ce5]
- Updated dependencies [2f1488e]
  - ai@6.0.289
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/togetherai@2.0.83

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/valibot@2.0.54

### Patch Changes

- @ai-sdk/provider-utils@4.0.53
## @ai-sdk/vercel@2.0.79

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/voyage@1.0.30

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/vue@3.0.289

### Patch Changes

- Updated dependencies [86f0089]
- Updated dependencies [da2e17b]
- Updated dependencies [f211ce5]
- Updated dependencies [2f1488e]
  - ai@6.0.289
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/xai@3.0.135

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53
## @ai-sdk/zai@2.0.5

### Patch Changes

- Updated dependencies [da2e17b]
  - @ai-sdk/provider@3.0.17
  - @ai-sdk/openai-compatible@2.0.77
  - @ai-sdk/provider-utils@4.0.53

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