Skip to content

feat(auto-model): route Auto Balanced Opus modes to Qwen3.7 Max#3568

Open
lambertjosh wants to merge 1 commit into
mainfrom
feat/auto-balanced-qwen37-max
Open

feat(auto-model): route Auto Balanced Opus modes to Qwen3.7 Max#3568
lambertjosh wants to merge 1 commit into
mainfrom
feat/auto-balanced-qwen37-max

Conversation

@lambertjosh
Copy link
Copy Markdown
Contributor

Summary

  • Align Auto Balanced mode routing with Auto Frontier's quality split: modes routed to Claude Opus in Frontier (claw, plan, general, architect, orchestrator, ask, and debug) now use qwen/qwen3.7-max in Balanced.
  • Keep the Sonnet-aligned Balanced paths (build, explore, code, and the no-mode default) on qwen/qwen3.6-plus, with a regression test guarding the tier relationship.

Verification

  • Not manually exercised; this is internal gateway model routing without a user-facing UI path.

Visual Changes

N/A

Reviewer Notes

  • Existing KiloClaw initial-setup routing and Alibaba-incompatible responses / messages fallback behavior remain unchanged.

@kilo-code-bot
Copy link
Copy Markdown
Contributor

kilo-code-bot Bot commented May 28, 2026

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Executive Summary

The Balanced tier mode routing is correctly restructured to mirror Frontier's quality split, with exhaustive type-safe mapping and a regression test guarding the tier relationship.

Files Reviewed (3 files)
  • apps/web/src/lib/ai-gateway/auto-model/index.ts
  • apps/web/src/lib/ai-gateway/auto-model/resolution.ts
  • apps/web/src/lib/ai-gateway/auto-model/index.test.ts
Other Observations (not in diff)

KILO_AUTO_BALANCED_MODEL in index.ts (unchanged lines ~155-170) still exposes a single set of pricing metadata derived from the lower-tier Qwen3.6 Plus model. After this PR, modes routed to qwen37_max_model incur roughly 5x higher token costs than the published metadata suggests. This is a pre-existing design concern that may warrant a follow-up to surface blended or worst-case pricing to users to avoid billing surprises.


Reviewed by claude-sonnet-4.6 · 721,098 tokens

Review guidance: REVIEW.md from base branch main

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.

1 participant