Skip to content

Fix model-translation gaps G1-G5 (effort ladders, -max rewrite, dead mimo-pro entry, muse fallback, snapshot re-sync) - #141

Merged
trefeon merged 12 commits into
mainfrom
model-translation-gaps
Aug 20, 2026
Merged

Fix model-translation gaps G1-G5 (effort ladders, -max rewrite, dead mimo-pro entry, muse fallback, snapshot re-sync)#141
trefeon merged 12 commits into
mainfrom
model-translation-gaps

Conversation

@trefeon

@trefeon trefeon commented Aug 19, 2026

Copy link
Copy Markdown
Owner

Description

Closes the model-translation gaps found in the research guide docs/guides/model-translation-layer.md (the per-model reference for the 15-model /v1/models catalog). Five correctness fixes plus docs:

  • G2 — isDeepSeekModel now matches the provisioned -max variants, so deepseek-v4-pro-max / deepseek-v4-flash-max get the medium-to-high reasoning rewrite and prompt-cache hints.
  • G1 — the effort table gains the three -max rows (same ladders as their base models).
  • G4 — removes the dead mimo/mimo-v2.5-pro effort entry (removed from FreeBuff free mode 2026-08-04; paid/BYOK only).
  • G3 — muse-spark now defaults to a deepseek-v4-pro fallback (mirrors MUSE_SPARK_FALLBACK_MODEL_ID); the FALLBACK_MODEL comments now state the true trigger (queue-wait only, never 429 quota exhaustion).
  • G5 — re-synced the pinned registry mirror LIMITED constants to mimo-only.

G6 and G7 are documented non-fixes: the gemini flash-lite rows are helper-subagent models (gated via MODELS_ALLOW), and the parser's unquoted-key skip is upstream-identical port parity.

Type of change

  • Bug fix
  • New feature
  • Refactor / internal
  • Docs only

Checklist

  • Code builds: go build ./...
  • go vet ./... clean
  • Tests pass: go test ./... (hermetic; race runs in CI — cgo unavailable locally)
  • Tests added/updated for the change
  • Public docs updated if behavior or config changed (model-translation-layer.md, 9router-integration.md)
  • No secrets: no real tokens, .env, or config.json content

Notes

  • Full hermetic suite 20/20 green on the branch head (0d4bd33); gofmt clean.
  • docs/guides/model-translation-layer.md is new in this branch and documents the full model-ID translation pipeline (aliases, fallbacks, -max variants, effort ladders, agent roots).

trefeon added 11 commits August 19, 2026 23:12
Shipped 0.0.150 binary carries CODEBUFF_CLI_VERSION 0.0.150
(extracted .bun section on vps-sg). Ads header UA was 0.0.149.
Body UA stays Chrome/124: the source's 151.0.0.0 change postdates
the shipped binary.
Quota ladder (premium 4 / limited 3, level ladder to 7; live server
still 6/day for limited), SG elevated (\$5 spend) / CN restricted-only,
MiMo V25 sole limited model, DeepSeek peak hours, Chrome-124 binary
truth (151 source drift), Freebuff-CLI/0.0.150, zeroclick ads network,
session standing/level/peak fields. Stale 6/day and 5-concurrent
claims fixed.
Copilot AI balanced review requested due to automatic review settings August 19, 2026 19:20

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

@trefeon
trefeon merged commit 7c390ef into main Aug 20, 2026
5 checks passed
@trefeon
trefeon deleted the model-translation-gaps branch August 20, 2026 05:25
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