chore(models): add fable tier and refresh Zen alias maps - #3
Conversation
The logical tier vocabulary gains fable (frontier tier above opus; Claude Fable 5 in Claude Code, GA 2026-06-09). Smoke-test alias maps refreshed against the live OpenCode Zen catalog: deepseek-v4-pro was removed from Zen, so kimi-k2.6 (fable) and glm-5.1 (opus) replace it; contract test now guards against deepseek-v4-pro reappearing. README documents recommended per-harness alias maps for Claude Code, Codex, Gemini CLI, OpenCode, Pi, Grok Build, and Antigravity. Evidence: bun test 217 pass / 0 fail.
|
Caution Review failedThe pull request is closed. ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (2)
📝 WalkthroughWalkthroughAdds a ChangesModel Tier System and Catalog Updates
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@README.md`:
- Around line 279-291: Update the per-harness alias table in README to match
validated identifiers: ensure OpenCode/Pi rows exactly match
approvedModelAliasMaps used in scripts/headless-smoke.ts (including removal of
deepseek-v4-pro and replacing nemotron-3-super-free with nemotron-3-ultra-free),
annotate that Codex "effort xhigh/high" is a host-only reasoning-effort (move it
to a footnote or separate column so it’s not part of the model id), and add a
short caveat for unverified harness rows (Grok/Antigravity) noting they may
drift and require external validation of tokens like grok-build vs
grok-build-0.1 and antigravity-preview-05-2026.
In `@scripts/headless-smoke.ts`:
- Around line 57-63: Update the comment and any tests referencing the OpenCode
Zen catalog to use the catalog's raw id + owned_by values rather than the
constructed config prefixes: adjust wording around approvedModelAliasMaps (the
opencode map entries for fable and opus) to state that catalog ids are
kimi-k2.6, glm-5.1, qwen3.6-plus, and deepseek-v4-flash-free and that
deepseek-v4-pro is not present; clarify that the opencode-go/... strings in
approvedModelAliasMaps are our internal/constructed names and not the catalog id
field, and ensure any assertions or comments compare owned_by + id (e.g.,
owned_by: "opencode" with id: "kimi-k2.6") rather than expecting the prefixed
form.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 307a3b35-ed32-444d-b9b1-69ef55913096
📒 Files selected for processing (3)
README.mdscripts/headless-smoke.tstests/headless-smoke-contract.test.ts
What
fableto the logical tier vocabulary (frontier tier aboveopus; Claude Fable 5 in Claude Code, GA 2026-06-09). No core code change needed:model-policy.tsis alias-generic.approvedModelAliasMaps+approvedPiFallbackModelsinscripts/headless-smoke.tsagainst the live OpenCode Zen catalog (2026-06-09):deepseek-v4-prowas removed from Zen:opencode-go/kimi-k2.6takesfable,opencode-go/glm-5.1takesopusdeepseek-v4-proandclaude-fablefrom reappearing in the mapsEvidence
Summary by CodeRabbit
Documentation
Updates
Tests