Atlas

Benchmarks

← All benchmarks

GDPx: Quant Trading

Finance & Accounting · 2026-11-01

GDPx's Quant Trading domain: 10 tasks from Third Moment on finding edge that survives out of sample on the full IEX order-book tape, graded on held-out daily P&L. Scores are pass@1 over three attempts per task. A component of the GDPx score, which weights its 11 domains equally.

Top models (higher is better)

ModelScore
Claude Fable 5.128.6
Claude Opus 5.527.4
GPT-6.1 Terra20.8
GPT-6 Astra20.2
Muse Wonder Full18.3
GPT-6.1 Sol17.1
MiniMax M3.1 Flash15.5
Muse Wonder14.8
Gemini 4 Carbon13.1
Claude Sonnet 5.513.0
GLM-5.311.0
Grok 4.79.1
Kimi K39.0
Claude Haiku 5.56.7
Gemini 4 Argon6.2
GPT-6 Luna5.7
Muse Spark Open0.9
Loading Atlas data…