fix(video): use GA Veo 3.1 model IDs on Vertex AI

[smakosh]

Summary

• Google deprecated the veo-3.1-generate-preview and veo-3.1-fast-generate-preview model IDs on Vertex AI on 2026-04-02, causing the playground errors Publisher Model ... was not found and Permission 'aiplatform.endpoints.predict' denied.
• Update the upstream modelName for the google-vertex provider to the GA IDs veo-3.1-generate-001 and veo-3.1-fast-generate-001.
• Keep the public-facing gateway model IDs (veo-3.1-generate-preview / veo-3.1-fast-generate-preview) unchanged so existing API consumers are not broken.
• Update the internal modelName comparison in the gateway and playground that gates reference-image support.

Test plan

• Generate a video with Veo 3.1 in the playground (no inputs)
• Generate a video with Veo 3.1 Fast in the playground (no inputs)
• Generate a video with Veo 3.1 using a reference image (verify gating still allows it)
• Confirm avalanche/veo-3.1-* routes still work

🤖 Generated with Claude Code

Summary by CodeRabbit

• Updates
  • Updated Veo 3.1 model identifiers from preview to production variants.
  • Improved handling of reference-image video requests for Google Vertex integrations to ensure correct model selection and logging.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 365865a1-bbf9-49d5-bab1-fbb817f28fe9

📥 Commits

Reviewing files that changed from the base of the PR and between d898ae2 and 94763e9.

📒 Files selected for processing (1)

• apps/gateway/src/videos/videos.spec.ts

---

Walkthrough

Switches Google Vertex Veo 3.1 mappings from preview to production model identifiers (veo-3.1-generate-001, veo-3.1-fast-generate-001), removes "preview" wording in model descriptions, updates reference-mode eligibility checks, and adjusts tests to expect the new identifiers.

Changes

Cohort / File(s)	Summary
Model Definitions packages/models/src/models/google.ts	Changed google-vertex modelName entries to veo-3.1-generate-001 and veo-3.1-fast-generate-001; removed "preview" wording from the corresponding model descriptions.
Feature Eligibility Logic apps/gateway/src/videos/videos.ts, apps/playground/src/lib/video-gen.ts	Updated reference-image/video eligibility logic to gate on veo-3.1-generate-001 instead of veo-3.1-generate-preview.
Tests apps/gateway/src/videos/videos.spec.ts	Updated test expectations to assert veo-3.1-generate-001 as the used model mapping in jobs and logs.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely summarizes the main change: updating Google Vertex AI to use GA Veo 3.1 model IDs instead of preview versions.
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.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ 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/veo-3-1-vertex-model-ids

---

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 and usage tips.}

[coderabbitai]

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

apps/gateway/src/videos/videos.ts (1)

778-794: ⚠️ Potential issue | 🟠 Major

Test suite assertions need updates for internal model name change.

The code change in lines 778–794 is correct: the internal gate now checks provider.modelName against the GA veo-3.1-generate-001, and user-facing error messages correctly retain the *-preview slugs. However, the extensive test suite in apps/gateway/src/videos/videos.spec.ts contains 40+ explicit assertions and fixtures that still expect the old preview model IDs.

Specifically, these need updating:

• usedModel field assertions (e.g., lines 476, 516, 823, 973, 1155, 1248) currently expect "veo-3.1-generate-preview" but will receive the new GA ID
• Test request bodies (lines 64, 103, 177, 271, etc.) should be reviewed to ensure they match the new internal flow
• Mock server fallback in apps/gateway/src/test-utils/mock-openai-server.ts:1364 still defaults to "veo-3.1-generate-preview" and may need alignment

Update all test expectations to match the internal model ID that will now flow through the system.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/gateway/src/videos/videos.ts` around lines 778 - 794, Update the test
expectations and fixtures so the internal model ID flowing through tests is the
GA name "veo-3.1-generate-001" while user-facing error strings remain the
preview slug; specifically, change all assertions that check the usedModel field
(e.g., any expectations on usedModel currently asserting
"veo-3.1-generate-preview") to expect "veo-3.1-generate-001", update test
request bodies/fixtures that construct requests to reflect the new internal
model name, and update the mock server fallback in the mock-openai-server module
(the default model string) so it returns "veo-3.1-generate-001" instead of
"veo-3.1-generate-preview". Ensure no user-facing error message strings are
altered.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Outside diff comments:
In `@apps/gateway/src/videos/videos.ts`:
- Around line 778-794: Update the test expectations and fixtures so the internal
model ID flowing through tests is the GA name "veo-3.1-generate-001" while
user-facing error strings remain the preview slug; specifically, change all
assertions that check the usedModel field (e.g., any expectations on usedModel
currently asserting "veo-3.1-generate-preview") to expect
"veo-3.1-generate-001", update test request bodies/fixtures that construct
requests to reflect the new internal model name, and update the mock server
fallback in the mock-openai-server module (the default model string) so it
returns "veo-3.1-generate-001" instead of "veo-3.1-generate-preview". Ensure no
user-facing error message strings are altered.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: be1d64b2-3221-4a4d-9ed2-b09efe4fedc3

📥 Commits

Reviewing files that changed from the base of the PR and between a49b8de and d898ae2.

📒 Files selected for processing (3)

• apps/gateway/src/videos/videos.ts
• apps/playground/src/lib/video-gen.ts
• packages/models/src/models/google.ts