Atlas

Benchmarks

← All benchmarks

BLXBench Refactoring category

Code · 2026-04-25

Refactoring category component of the BLXBench public leaderboard. The source describes this category as code transformation while preserving behavior and intent; scores are category score percentages.

Top models (higher is better)

ModelScore
North Mini Code69.1
GPT-5.3-Codex67.1
Qwen3.7-Max66.7
MiMo-V2.566.0
GPT-5.565.2
Nemotron 3 Super 120B A12B64.7
GLM-5.164.3
Granite 4.1 8B63.4
MiniMax M362.5
Mistral Small 462.5
DeepSeek-V4-Pro61.9
Mistral Medium 3.561.2
DeepSeek-V4-Flash61.0
Grok Build 0.160.1
GLM-5.259.9
Nemotron 3 Nano 30B A3B59.5
Kimi K2.659.3
Qwen3.6-Flash-2026-04-1659.0
MiMo-V2.5-Pro55.8
Grok 4.355.6
Qwen3.7-Plus55.1
Ring 2.6 1T54.7
CoBuddy51.0
Nemotron 3 Nano Omni 30B A3B50.8
Opus 4.846.4
Gemini 3.1 Flash-Lite43.8
Claude Fable 542.5
Kimi K2.7 Code37.3
Opus 4.736.8
MiniMax M2.728.8
Gemini 3.5 Flash25.0
Step 3.7 Flash17.7
Loading Atlas data…