Atlas

Benchmarks

← All benchmarks

Buyout Game Bradley-Terry Rating

Games · 2026-03-30

Buyout Game is a game-theoretic benchmark for model negotiation and purchasing decisions. The public leaderboard reports Bradley-Terry ratings over completed games.

Top models (higher is better)

ModelScore
GPT-5.51975
GPT-5.41961
Opus 4.71878
GLM-51792
Opus 4.61760
DeepSeek-V4-Pro1682
Gemini 3.5 Flash1667
Sonnet 4.61656
Hy3 preview1621
Gemini 3.1 Flash-Lite Preview1615
GLM-5.11605
Gemini 3.1 Pro Preview1564
DeepSeek-V4-Flash1556
Grok 4.201549
Grok 4.31525
Gemma 4 31B IT1522
MiMo-V2.5-Pro1517
Kimi K2.61468
Kimi K2.51462
Qwen3.6-Max-Preview1459
MiMo-V2-Pro1428
Step 3.5 Flash1416
Doubao Seed 2.0 Pro1405
DeepSeek-V3.21381
MiniMax M2.71374
Mistral Medium 3.51332
Qwen3.5 397B A17B1310
ERNIE 5.11211
Mistral Large 3 675B Instruct 25121206
Llama 4 Maverick1128
ERNIE 5.01043
gpt-oss-120b997
Loading Atlas data…