fix(zoo-gateway): enable image attach when model supports vision

[JamesRobert20]

Summary

• Gate image attachment on whether the selected Zoo Gateway / Vercel AI Gateway model actually supports vision
• Read that capability from live /models tags (including vision`) instead of only a hardcoded allowlist that drifts out of date
• Keep the allowlist as a fallback when tags are missing from older catalog responses

Why

Image attach in chat is controlled by supportsImages. For Zoo Gateway (and Vercel AI Gateway), that flag came from a static model-ID allowlist, so vision-capable models that were not listed still had image input disabled. This makes capability detection follow the models API.

Test plan

• pnpm exec vitest run api/providers/fetchers/__tests__/vercel-ai-gateway.spec.ts api/providers/fetchers/__tests__/zoo-gateway.spec.ts (from src/)
• Zoo Gateway signed in + vision-tagged model → image attach enabled
• Non-vision tagged model → image attach disabled
• Catalog response without tags → allowlisted vision models still enable attach

Summary by CodeRabbit

• New Features

  • Added support for additional Anthropic and OpenAI models through the Vercel AI Gateway.
  • Improved image-capability detection using live model tags, with fallback support for existing model capabilities.
  • Added vision capability detection for tagged Zoo Gateway models.

• Bug Fixes

  • Improved accuracy when identifying models that support image inputs, including models without capability tags.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 966e9626-8274-4638-a564-de0049f550be

📥 Commits

Reviewing files that changed from the base of the PR and between 1d7703c and 42c4b8a.

📒 Files selected for processing (1)

• src/api/providers/fetchers/vercel-ai-gateway.ts

💤 Files with no reviewable changes (1)

• src/api/providers/fetchers/vercel-ai-gateway.ts

---

📝 Walkthrough

Walkthrough

The Vercel AI Gateway model schema now accepts capability tags, and image support detection prefers the vision tag before using static allowlists. Additional Anthropic and OpenAI vision-capable models were added, with tests covering tagged and fallback behavior.

Changes

Gateway capability detection

Layer / File(s)	Summary
Vision contract and allowlist expansion packages/types/src/providers/vercel-ai-gateway.ts, src/api/providers/fetchers/vercel-ai-gateway.ts	Vercel AI Gateway model entries now support optional tags, and the vision-and-tools allowlist includes additional Anthropic and OpenAI models.
Tag-based image support parsing src/api/providers/fetchers/vercel-ai-gateway.ts, src/api/providers/fetchers/__tests__/vercel-ai-gateway.spec.ts, src/api/providers/fetchers/__tests__/zoo-gateway.spec.ts	Image support is derived from vision tags when available and falls back to static allowlists otherwise; Vercel and Zoo Gateway tests cover the behavior.

Estimated code review effort: 2 (Simple) | ~10 minutes

Sequence Diagram(s)

sequenceDiagram
  participant VercelAIGateway
  participant ModelSchema
  participant ModelParser
  participant VisionAllowlists
  VercelAIGateway->>ModelSchema: Provide model data with optional tags
  ModelSchema->>ModelParser: Pass validated model
  ModelParser->>ModelParser: Check tags for vision
  ModelParser->>VisionAllowlists: Check fallback when tags are absent
  VisionAllowlists-->>ModelParser: Return vision capability result

Loading

Possibly related PRs

• Zoo-Code-Org/Zoo-Code#344: Also changes Vercel AI Gateway parsing behavior used by Zoo Gateway delegation.
• Zoo-Code-Org/Zoo-Code#555: Also updates the Vercel AI Gateway fetcher and model-specific parsing logic.

Suggested labels: awaiting-review

Suggested reviewers: taltas, navedmerchant, hannesrudolph, edelauna

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description covers summary, rationale, and tests, but it omits the required linked GitHub issue and several template sections.	Add a Related GitHub Issue section with Closes: #, and complete the checklist plus any required documentation, screenshots, and contact fields.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: enabling image attach for vision-capable Zoo Gateway models.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/zoo-gateway-vision-capabilities

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!