PACT Negotiation CMS
Agents · 2025-07-29
PACT Composite Model Score combines opponent-balanced performance with captured economic surplus using the benchmark's fixed alpha blend. Higher CMS indicates stronger negotiating performance in the public leaderboard.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5 | 69.0 |
| GPT-5.5 | 61.0 |
| Opus 4.8 | 56.0 |
| DeepSeek-V4-Pro | 54.0 |
| GLM-5.2 | 54.0 |
| Gemini 3.1 Pro Preview | 54.0 |
| Mistral Medium 3.5 | 52.0 |
| Gemma 4 31B IT | 50.0 |
| Sonnet 4.6 | 50.0 |
| Kimi K2.6 | 49.0 |
| GLM-5.1 | 49.0 |
| Hy3 preview-Base | 49.0 |
| gpt-oss-120b | 48.0 |
| Gemini 3.1 Flash-Lite Preview | 47.0 |
| Haiku 4.5 | 46.0 |
| Doubao Seed 2.0 Pro | 45.0 |
| Trinity Large Thinking | 44.0 |
| MiMo-V2.5-Pro | 44.0 |
| Qwen3.7-Max | 41.0 |
| Grok 4.3 | 40.0 |
| Step 3.7 Flash | 37.0 |
| DeepSeek-V4-Flash | 37.0 |
| ERNIE 5.1 | 32.0 |
| Nova Pro | 32.0 |
| Mistral Large 3 675B Instruct 2512 | 28.0 |
| Llama 4 Maverick | 26.0 |