Atlas

Benchmarks

← All benchmarks

CadQueryEval Corrected Geometry Hausdorff

Code · 2026-09-30

CadQueryEval final corrected-geometry release: percentage of 25 meshes passing the 1 mm 95th-percentile Hausdorff-distance threshold. Task 6 may match either reviewed reference interpretation.

Top models (higher is better)

ModelScore
Claude Opus 5.5100.0
Gemini 3.8 Flash100.0
GPT-5.6 Sol Pro100.0
GPT-6 Astra100.0
GPT-6 Sol100.0
GPT-6 Luna96.0
Gemini 3.1 Pro Preview92.0
GLM-5.3 Flash92.0
GPT-5.6 Luna Pro92.0
GPT-5.6 Sol92.0
Grok 4.792.0
Gemini 3.7 Flash88.0
GPT-5.6 Luna88.0
Kimi K388.0
Muse Spark 1.388.0
Claude Fable 584.0
GLM-5.384.0
GPT-5.6 Terra Pro84.0
Qwen3.7-Max84.0
Claude Fable 5.180.0
Opus 4.880.0
Claude Opus 580.0
MiMo-V2.6-Pro80.0
Opus 4.676.0
DeepSeek V4.1 Flash76.0
Gemini 3.5 Flash76.0
Gemini 3.6 Flash76.0
GPT-5.576.0
Grok 4.576.0
Grok 4.676.0
Sonnet 572.0
Opus 4.768.0
GPT-5.6 Terra68.0
Grok 4.368.0
MiMo-V2.6-Flash68.0
Muse Spark 1.168.0
Gemini 3 Pro Preview64.0
GLM-5.164.0
Kimi K2.664.0
Muse Spark 1.264.0
Loading Atlas data…