Skip to content

Remove MiMo V2.5 Pro and Gemini 3.1 Pro Preview models#183

Merged
chloeilabs merged 1 commit into
mainfrom
claude/epic-volta-l5fvot
Jun 15, 2026
Merged

Remove MiMo V2.5 Pro and Gemini 3.1 Pro Preview models#183
chloeilabs merged 1 commit into
mainfrom
claude/epic-volta-l5fvot

Conversation

@chloeilabs

Copy link
Copy Markdown
Owner

Summary

Completely removes MiMo V2.5 Pro (xiaomi/mimo-v2.5-pro) and Gemini 3.1 Pro Preview (google/gemini-3.1-pro-preview) from the app, leaving Qwen 3.7 Max and Kimi K2.6 as the only models. The chat model selector reads from the registry, so both disappear from the UI automatically.

Changes

Registry & prompt steering

  • models.ts — dropped both from AvailableModels, SUPPORTED_MODELS, MODEL_SELECTOR_MODELS, and ModelInfos.
  • agent-prompt-steering.ts — narrowed PromptProvider to "alibaba" | "moonshotai", removed the google/xiaomi provider overlays, and dropped the resolver branches.

Dead-code cleanup

  • Deleted ai-sdk-gateway-provider-options.ts — it existed solely to set Gemini's thinkingConfig knob (no other provider used it), so it became a pure no-op once Gemini was gone.
  • Removed the now-unused taskMode threading from agent-runtime.ts and agent-route.ts (it was only consumed by that Gemini-specific provider-options call). Task mode is still inferred and still drives the system-prompt overlay.
  • Fixed a stale Gemini reference in an agent-runtime-synthesis-gating.ts comment.

Tests & docs

  • Deleted the Gemini-only agent-runtime-task-mode-routing.test.mjs; updated the registry, prompt-steering, system-prompt, helper-behavior, payload, gateway-search, and stub tests. The registry test now asserts both ids are fully absent.
  • Updated CLAUDE.md (Model Registry, provider overlay list, task-mode note, file tree) and README.md.

Verification

All green locally:

  • pnpm typecheck
  • pnpm lint ✅ (--max-warnings=0)
  • pnpm format:check
  • pnpm test ✅ (178 tests)
  • pnpm build

https://claude.ai/code/session_01JzzEBjdPyv9GmBrz6PEeWP


Generated by Claude Code

Drop both models from the shared registry and clean up every reference:

- models.ts: remove from AvailableModels, SUPPORTED_MODELS,
  MODEL_SELECTOR_MODELS, and ModelInfos (Qwen 3.7 Max and Kimi K2.6 remain)
- agent-prompt-steering.ts: drop the google/xiaomi provider overlays,
  narrow PromptProvider, and remove the resolver branches
- Delete ai-sdk-gateway-provider-options.ts (it only configured Gemini's
  thinking knob) and strip its now-dead taskMode threading through
  agent-runtime and agent-route
- Update tests + docs; delete the Gemini-only provider-options test

https://claude.ai/code/session_01JzzEBjdPyv9GmBrz6PEeWP
@vercel

vercel Bot commented Jun 15, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
chloei Ready Ready Preview, Comment Jun 15, 2026 5:25am

@chloeilabs chloeilabs merged commit e67dc2d into main Jun 15, 2026
6 checks passed
@chloeilabs chloeilabs deleted the claude/epic-volta-l5fvot branch June 15, 2026 05:29
@chloeilabs chloeilabs mentioned this pull request Jun 15, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants