Atlas

Benchmarks

← All benchmarks

GDPx: Software Engineering

Code · 2026-11-01

GDPx's Software Engineering domain: 15 tasks from Datacurve on long-horizon feature work, debugging and migrations in large production codebases. 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 Sonnet 5.511.1
Gemini 4 Carbon11.1
GPT-6.1 Terra11.1
Claude Opus 5.58.9
Gemini 4 Argon8.9
GPT-6.1 Sol8.9
Claude Haiku 5.56.7
GPT-6 Astra6.7
Claude Fable 5.14.4
MiniMax M3.1 Flash2.2
Muse Wonder Full2.2
Muse Wonder2.2
GLM-5.30.0
GPT-6 Luna0.0
Grok 4.70.0
Kimi K30.0
Muse Spark Open0.0
Loading Atlas data…