Benchmarks · reasoning
Olam Poker Elo
Uses Olam Labs' Bradley-Terry Elo ratings from multi-agent social poker and excludes human players. Score: Poker Elo scaled expected win rate vs Claude Fable 5. Collected by Jake Boggs' benchmark dashboard.
34 models · 34 results · data as of 2026-10-29 · olamlabs.ai ↗
Leaderboard
Fable 5 max
All results
34 rows| # | Model | Effort | Olam Poker Elo |
|---|---|---|---|
| 1 | Fable 5 | max | 1553 |
| 2 | GPT-6 Astra | max | 1550 |
| 3 | GPT-6.1 Sol | max | 1543 |
| 4 | GPT-6 Sol | max | 1542 |
| 5 | GPT-5.6 Sol | max | 1534 |
| 6 | Sonnet 5.5 | max | 1526 |
| 7 | GPT-5.5 | xhigh | 1521 |
| 8 | GPT-6 Luna | max | 1521 |
| 9 | GLM 5.3 Flash | max | 1520 |
| 10 | Opus 4.8 | max | 1518 |
| 11 | Opus 5 | max | 1518 |
| 12 | DeepSeek V4.1 Flash | max | 1515 |
| 13 | GPT-5.6 Terra | max | 1511 |
| 14 | Sonnet 5 | max | 1509 |
| 15 | Grok 4.7 | max | 1508 |
| 16 | Opus 5.5 | max | 1506 |
| 17 | Muse Spark 1.1 | max | 1504 |
| 18 | DeepSeek V4 Pro | max | 1500 |
| 19 | Grok 4.6 | max | 1499 |
| 20 | Gemini 3.5 Flash | xhigh | 1496 |
| 21 | Gemini 3.1 Pro | high | 1495 |
| 22 | Kimi K3 | max | 1495 |
| 23 | Fable 5.1 | max | 1494 |
| 24 | Muse Spark 1.3 | max | 1494 |
| 25 | GLM 5.2 | max | 1492 |
| 26 | GLM 5.3 | max | 1492 |
| 27 | Gemini 3.6 Flash | xhigh | 1489 |
| 28 | Grok 4.5 | max | 1487 |
| 29 | GPT-5.6 Luna | max | 1486 |
| 30 | Gemini 3.8 Flash | max | 1479 |