Atlas

Benchmarks

← All benchmarks

FrontierCode 1.0 Extended - Pass Rate

Code · 2026-06-08

FrontierCode 1.0 Extended pass rate across the full set of 150 production-code tasks. A trial passes only when it satisfies every blocker criterion, approximating whether a repository maintainer would merge the patch.

Top models (higher is better)

ModelScore
Claude Fable 565.3
Opus 4.856.1
GPT-5.549.0
Opus 4.747.5
Kimi K2.7 Code43.7
GLM-5.242.8
Kimi K2.640.7
GPT-5.4 Mini39.9
Gemini 3.1 Pro Preview38.0
Sonnet 4.636.8
Kimi K2.525.0
Qwen3.6 Plus (2026-04-02)23.5
MiniMax M2.722.0
SWE-1.620.0
MiniMax M2.517.3
Gemini 3.1 Flash-Lite Preview16.2
Loading Atlas data…