Atlas

Benchmarks

← All benchmarks

Vending-Bench (minimum units sold)

Games · 2025-02-20

Minimum units sold across repeated runs in the original Vending-Bench long-horizon vending business simulation.

Top models (higher is better)

ModelScore
Grok 43515
Gemini 3 Pro Preview3165
Sonnet 4.51907
GPT-51798
Opus 41218
GPT-5.11166
Human (Vending-Bench)344
gpt-oss-120b338
o3331
Gemini 2.5 Pro Preview 03-25313
Gemini 2.0 Pro Experimental118
GPT-4o108
Gemini Robotics-ER 1.575.0
GPT-5 Mini72.0
GPT-4o Mini65.0
Claude 3.5 Sonnet (Oct 2024)0.0
Qwen3-235B-A22B0.0
Sonnet 40.0
Claude 3.7 Sonnet0.0
o3-mini0.0
Gemini 1.5 Pro0.0
Gemini 1.5 Flash0.0
Haiku 3.50.0
Gemini 2.0 Flash0.0
Llama 4 Maverick0.0
Kimi K2 Thinking0.0
Loading Atlas data…