Atlas

Benchmarks

← All benchmarks

FrontierSWE: Lua Native Compiler (Best@5)

Code · 2026-04-16

Build a real AOT compiler from Lua 5.4 source to native x86-64 ELF. FrontierSWE Best@5 is the highest percentage of verifier tests passed among up to five independent long-horizon agent runs.

Top models (higher is better)

ModelScore
Claude Fable 599.0
GLM-5.290.0
Opus 4.789.0
Opus 4.889.0
GPT-5.488.0
GPT-5.588.0
Opus 4.685.0
Composer 2.530.0
DeepSeek-V4-Pro26.0
Gemini 3.1 Pro Preview20.0
Kimi K2.52.8
GLM-5.10.0
Grok 4.50.0
Kimi K2.60.0
Qwen3.6 Plus (2026-04-02)0.0
Loading Atlas data…