Atlas

Benchmarks

← All benchmarks

SnorkelManufacturing

Professional Work

Pass@1 on 200 simulated discrete-parts manufacturing tasks combining CAD reasoning, engineering evidence, production and quality decisions, safety constraints and controlled state changes. Native effort and agent scaffold are not disclosed.

Top models (higher is better)

ModelScore
Grok 4.616.4
Grok 4.712.1
GLM-5.311.9
Claude Fable 5.111.3
Muse Spark 1.310.5
Claude Opus 5.510.4
Kimi K310.2
GPT-6 Astra10.0
Claude Opus 59.4
Gemini 3.8 Flash7.1
Loading Atlas data…