Skip to content

feat(auto): replace kilo-auto/small backing with Gemma 4#2502

Merged
chrarnoldus merged 3 commits into
mainfrom
feat/kilo-auto-small-gemma-4
Apr 17, 2026
Merged

feat(auto): replace kilo-auto/small backing with Gemma 4#2502
chrarnoldus merged 3 commits into
mainfrom
feat/kilo-auto-small-gemma-4

Conversation

@kilo-code-bot

@kilo-code-bot kilo-code-bot Bot commented Apr 16, 2026

Copy link
Copy Markdown
Contributor

Summary

Replace the backing models for kilo-auto/small:

  • Paid users: openai/gpt-5-nanogoogle/gemma-4-31b-it
  • Free users: openai/gpt-oss-20b:freegoogle/gemma-4-26b-a4b-it:free

Key changes:

  • Context window increased from 131K to 262K tokens
  • Image/vision support enabled
  • Preferred inference provider set to Novita for Gemma models
  • gpt-oss-20b:free status set to disabled (retained in kiloExclusiveModels per deprecation policy)
  • Test snapshot updated to reflect new capabilities

This PR was authored by anthropic/claude-opus-4.6, a premium model.

Verification

  • pnpm typecheck passes
  • pnpm format applied
  • Approved snapshot updated for kilo-auto/small model properties

Visual Changes

N/A

Reviewer Notes

  • The gpt_oss_20b_free_model is set to disabled rather than removed, following the existing deprecation policy in models.ts
  • The free-tier Gemma model is defined as a KiloExclusiveModel with status: 'hidden' (same pattern as the old gpt-oss-20b:free)
  • isGemmaModel() check added to getPreferredProviderOrder() to route all Gemma models through Novita

Switch kilo-auto/small from GPT-5 Nano / gpt-oss-20b:free to
google/gemma-4-31b-it (paid) and google/gemma-4-26b-a4b-it (free).

- Context window increased from 131K to 262K tokens
- Image support enabled
- Preferred provider set to Novita for both models
- gpt-oss-20b:free model status set to disabled
Comment thread apps/web/src/lib/ai-gateway/providers/index.ts
@kilo-code-bot

kilo-code-bot Bot commented Apr 16, 2026

Copy link
Copy Markdown
Contributor Author

Code Review Summary

Status: 1 Issue Found | Recommendation: Address before merge

Overview

Severity Count
CRITICAL 0
WARNING 1
SUGGESTION 0
Issue Details (click to expand)

WARNING

File Line Issue
apps/web/src/lib/ai-gateway/providers/index.ts 282 Gemma requests can still be routed onto Vercel while carrying a Novita-only provider preference, which can break kilo-auto/small routing.

Fix these issues in Kilo Cloud

Other Observations (not in diff)

Issues found in unchanged code that cannot receive inline comments:

File Line Issue
apps/web/src/lib/ai-gateway/providers/vercel/index.ts 90 The latest commit removes the Gemma Vercel-routing guard, reintroducing the existing warning above, but this file is no longer present in the current gh pr diff output so it cannot receive a new inline comment.
Files Reviewed (4 files)
  • apps/web/src/lib/ai-gateway/models.ts - 0 issues
  • apps/web/src/lib/ai-gateway/providers/google.ts - 0 issues
  • apps/web/src/lib/ai-gateway/providers/openai.ts - 0 issues
  • apps/web/src/lib/ai-gateway/providers/vercel/index.ts - 1 issue

Reviewed by gpt-5.4-20260305 · 913,754 tokens

kilo-code-bot Bot added 2 commits April 16, 2026 12:19
Novita has no Vercel inference provider mapping, so Gemma requests
routed to Vercel would fail. Exclude Gemma from the Vercel
randomization to keep all requests on OpenRouter.
…date Gemma description

- Revert Gemma exclusion from Vercel routing (unnecessary)
- Remove gpt_oss_20b_free_model definition and all imports
- Update Gemma 4 26B description from OpenRouter API metadata
@chrarnoldus chrarnoldus requested review from joshavant and lambertjosh and removed request for joshavant April 16, 2026 15:45
@chrarnoldus chrarnoldus merged commit 416ca73 into main Apr 17, 2026
15 checks passed
@chrarnoldus chrarnoldus deleted the feat/kilo-auto-small-gemma-4 branch April 17, 2026 07:46
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