Automated changes by
[create-pull-request](https://github.com/peter-evans/create-pull-request)
GitHub action

Co-authored-by: loreto <[email protected]>

Automated changes by
[create-pull-request](https://github.com/peter-evans/create-pull-request)
GitHub action

Co-authored-by: loreto <[email protected]>

Automated changes by
[create-pull-request](https://github.com/peter-evans/create-pull-request)
GitHub action

Co-authored-by: loreto <[email protected]>

## Summary
Adds several TODOs of changes I'm considering to the AI SDK, and links
to relevant codebases (in MCP or Vercel's version, for example).

Just want to get these in quickly before I start making changes.

## How was it tested?
N/A

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

Automated changes by
[create-pull-request](https://github.com/peter-evans/create-pull-request)
GitHub action

Co-authored-by: loreto <[email protected]>

## Summary
Simplify the API by removing the need for "RedactedReasoning". Instead
we have a single Reasoning block, and if there's fields related to a
redaction, we keep them in the provider metadata.

This follows the recent changes in the Vercel SDK.

## How was it tested?
Updated tests and ran them.

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

## Summary
The AI SDK has types defined for image models and embedding models ...
but we've never really used those or implemented providers for those
models. This PR moves those types to internal, so that end-users don't
start depending on those types until we test them out and are ready to
make them public.

## How was it tested?
Ran tests.

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

## Summary

Ran into panic with this:
```
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x50 pc=0x11a77a0]

goroutine 561 [running]:
go.jetify.com/ai/api.(*Response).GetProviderMetadata(0x0?)
	/home/runner/go/pkg/mod/go.jetify.com/[email protected]/api/llm_language_model.go:93
2025-10-17 03:26:09.667
go.jetify.com/ai/api.GetMetadata[...]({0x20f03cb, 0x9}, {0x23e0540, 0x0?})
2025-10-17 03:26:09.667
	/home/runner/go/pkg/mod/go.jetify.com/[email protected]/api/llm_provider_metadata.go:95 +0x5f
2025-10-17 03:26:09.667
go.jetify.com/ai/provider/anthropic/codec.GetMetadata(...)
2025-10-17 03:26:09.667
	/home/runner/go/pkg/mod/go.jetify.com/[email protected]/provider/anthropic/codec/metadata.go:24
2025-10-17 03:26:09.667
```

## How was it tested?

have not tested

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

## Summary

This reverts commit c6b33e126335c366c106eb5f28fcafab2887ae37.
That was a silly no-op

## How was it tested?

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

…etadata (#570)

## Summary

This should fix the nil pointer seg-fault hit in #568

## How was it tested?

Exercised product code that hit this code path, and verified that the
anthropic code.GetMetadata() call was able to return successfully

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

## Summary
- Pulls out tool IDs and names into constants, to avoid using string
literals in multiple places
- Adds ComputerUsePreview as an available model
- Fixes bug were ComputerUse ToolResults were not being encoded properly
because we were comparing `toolCallID` (e.g. `cu_deadbeef`) against the
tool name (`openai.computer_use_preview`)

NOTE: we are setting the Tool ID into `api.ToolCallBlock.ToolName` and
`api.ToolResultBlock.ToolName`. This seems like a bug. I didn't fix it
in this PR; will fix in the future. I'm not sure if it makes more sense
to expose the ToolID or the ToolName. I think ToolID but let me know if
you have opinions.

## How was it tested?
CICD

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

Automated changes by
[create-pull-request](https://github.com/peter-evans/create-pull-request)
GitHub action

Co-authored-by: loreto <[email protected]>

## Summary
Populates `ResponseInfo` in the response object for openai calls.

## How was it tested?
CICD

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

Automated changes by
[create-pull-request](https://github.com/peter-evans/create-pull-request)
GitHub action

Co-authored-by: ipince <[email protected]>

## Summary
CUA models require auto-truncation.

I also set `isReasoning` and `systemMessageMode` to match what Vercel's
AISDK sets for them, but honestly I'm not sure how those fields are
used... The inference calls worked just fine whether the roles were
"system" or "developer".

Also, it seems that CUA is indeed a "reasoning model", because we do get
reasoning output from it, even if it's not listed as a reasoning model
in OpenAI's docs. But, OpenAI's docs also say that reasoning models do
_not_ support temperature, but we do set temperature (after
loop-detection) and it does seem to work... so I'm a bit confused here.
I might have to change some of the AISDK code that currently enforces
that temperature is NOT set for reasoning models.

## How was it tested?
Ran `testpilot test` while pulling these local changes.

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

## Summary
TSIA

## How was it tested?
Unit test; running testpitot locally

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

## Summary
Sometimes the model does return an empty reasoning block... Let's let it
through instead of failing to decode.

Example:

https://platform.openai.com/logs/resp_02474882b463981300691523a7eb9481938b7823951cfe62ab

I'll add the raw response in a comment to this PR

## How was it tested?
Ran testpilot locally.

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).

## Summary
We need it so that we can specify whether the tool schema validation
should be strict or not.

## How was it tested?
Used with testpilot and its custom tools. I'll send a PR on axiom to
show its usage too.

## Community Contribution License

All community contributions in this pull request are licensed to the
project maintainers under the terms of the [Apache 2
License](https://www.apache.org/licenses/LICENSE-2.0).

By creating this pull request I represent that I have the right to
license the contributions to the project maintainers under the Apache 2
License as stated in the [Community Contribution
License](https://github.com/jetify-com/opensource/blob/main/CONTRIBUTING.md#community-contribution-license).