Name: BAZAAR TrueSkill Mu
Published: 2025-07-21

BAZAAR TrueSkill Mu

Agents · 2025-07-21

BAZAAR evaluates language models and baselines in a competitive double-auction market with private values and incomplete information. The public leaderboard ranks participants by multi-pass TrueSkill mu.

Top models (higher is better)

Model	Score
o3	7.4
Gemini 2.5 Pro	7.2
Gemini 2.5 Flash	7.2
Sonnet 4	7.1
Grok 4	6.9
Opus 4	6.9
O4 Mini	6.8
DeepSeek-R1-0528	6.5
Qwen3-235B-A22B	6.5
ChatGPT-4o Latest observed 2025-03-27	5.9
Llama 4 Maverick	5.5
DeepSeek-V3-0324	5.2
Nova Pro	5.2
Haiku 3.5	5.0
Qwen3-30B-A3B	5.0
Llama 4 Scout	4.9
GPT-4o Mini	4.8
Mistral Small 3.2 24B Instruct 2506	4.8
Mistral Medium 3	4.7
MiniMax-Text-01	4.6
Phi-4	4.4
Kimi K2 Instruct	4.4
ERNIE 4.5 300B A47B	4.3
Gemma 3 27B IT	4.2