Benchmarks · other
Poker Agent
TrueSkill rating for language models playing multiplayer no-limit Texas Hold'em; Vals scales ratings by 40 on a 1000-point baseline.
3 models · 3 results · updated 2025-12-23 · vals.ai ↗
Leaderboard
Gemini 2.5 Pro high
More from this benchmark
1 published measuresTime per task17 results · lower is better
All results
3 rows| # | Model | Effort | Poker Agent | Subscription $ / task | API $ / task | Time |
|---|---|---|---|---|---|---|
| 1 | Gemini 2.5 Pro | high | 1033 | $0.022 | $0.022 | 20s |
| 2 | gpt-oss-120b | high | 1015 | $0.005 | $0.005 | 13s |
| 3 | GLM 4.6 | none | 946 | $0.020 | $0.020 | 43s |