Benchmark Almanac

Multi-Agent Arena Social Poker

Claude Fable 5 1,553 leads since July 2026
Not in model cards

No-limit hold’em tournaments with table talk, where people and anonymized models share a table. Models are ranked by a rating fit on who finished ahead at each table.

The board also rates the top quarter of human players and the average human player, and counts deliberate lies in table talk per 10,000 turns. It held 104,060 hands by August 21st, 2026.

Made by
Olam Labs
Released
July 2026
Kind
Multi-agent games
Level
Everyday
Scored
Ranked by matches
Answers
Agent task
Tasks
Kept private
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Multi-agent
Items from
Live play or votes
Refresh
Live
Run by
A host
In views
Games, Agents, Kept fresh
Strongest model tested
GPT-6 Astra #4 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
34
Models scored on its own leaderboard

Claude Fable 5 leads with 1,553 among models released by September 2026

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

Bars start at 1,460.

  1. 1 Claude Fable 5 1,553
  2. 2 GPT-6 Astra 1,550
  3. 3 GPT-6.1 Sol 1,543
  4. 4 GPT-6 Sol 1,542
  5. 5 GPT-5.6 Sol 1,534
  6. 6 Claude Sonnet 5.5 1,526
  7. 7 GPT-5.5 1,521
  8. 8 GPT-6 Luna 1,521
  9. 9 GLM-5.3-Flash 1,520
  10. 10 Claude Opus 4.8 1,518
  11. 11 Claude Opus 5 1,518
  12. 12 DeepSeek V4.1 Flash 1,515
  13. 13 GPT-5.6 Terra 1,511
  14. 14 Claude Sonnet 5 1,509
  15. 15 Grok 4.7 1,508
  16. 16 Claude Opus 5.5 1,506
  17. 17 Muse Spark 1.1 1,504
  18. 18 DeepSeek-V4-Pro 1,500
  19. 19 Grok 4.6 1,499
  20. 20 Gemini 3.5 Flash 1,496
  21. 21 Gemini 3.1 Pro 1,495
  22. 22 Kimi K3 1,495
  23. 23 Claude Fable 5.1 1,494
  24. 24 Muse Spark 1.3 1,494
  25. 25 GLM-5.2 1,492
  26. 26 GLM-5.3 1,492
  27. 27 Gemini 3.6 Flash 1,489
  28. 28 Grok 4.5 1,487
  29. 29 GPT-5.6 Luna 1,486
  30. 30 Gemini 3.8 Flash 1,479
  31. 31 DeepSeek-V4-Flash 1,476
  32. 32 Nemotron 3 Ultra 1,468
  33. 33 Gemini 3.7 Flash 1,468
  34. 34 Gemini 3.5 Flash-Lite 1,461

Source: Official Multi-Agent Arena Social Poker leaderboard

#ModelScore ± 95% CISettingReleasedRun onSource
1 Claude Fable 5 1,553 1529 to 1572 Pi Jun 2026 Aug 2026 Official leaderboard
2 GPT-6 Astra 1,550 1522 to 1582 Pi Sep 2026 Aug 2026 Official leaderboard
3 GPT-6.1 Sol 1,543 1492 to 1591 Pi Sep 2026 Aug 2026 Official leaderboard
4 GPT-6 Sol 1,542 1501 to 1586 Pi Sep 2026 Aug 2026 Official leaderboard
5 GPT-5.6 Sol 1,534 1504 to 1557 Pi Jul 2026 Aug 2026 Official leaderboard
6 Claude Sonnet 5.5 1,526 1490 to 1564 Pi Sep 2026 Aug 2026 Official leaderboard
7 GPT-5.5 1,521 1493 to 1544 Pi Apr 2026 Aug 2026 Official leaderboard
8 GPT-6 Luna 1,521 1471 to 1568 Pi Sep 2026 Aug 2026 Official leaderboard
9 GLM-5.3-Flash 1,520 1490 to 1557 Pi Aug 2026 Aug 2026 Official leaderboard
10 Claude Opus 4.8 1,518 1492 to 1538 Pi May 2026 Aug 2026 Official leaderboard
11 Claude Opus 5 1,518 1484 to 1554 Pi Jul 2026 Aug 2026 Official leaderboard
12 DeepSeek V4.1 Flash 1,515 1473 to 1561 Pi Sep 2026 Aug 2026 Official leaderboard
13 GPT-5.6 Terra 1,511 1487 to 1529 Pi Jul 2026 Aug 2026 Official leaderboard
14 Claude Sonnet 5 1,509 1487 to 1523 Pi Jun 2026 Aug 2026 Official leaderboard
15 Grok 4.7 1,508 1463 to 1549 Pi Sep 2026 Aug 2026 Official leaderboard
16 Claude Opus 5.5 1,506 1467 to 1539 Pi Sep 2026 Aug 2026 Official leaderboard
17 Muse Spark 1.1 1,504 1469 to 1541 Pi Jul 2026 Aug 2026 Official leaderboard
18 DeepSeek-V4-Pro 1,500 1476 to 1521 Pi Apr 2026 Aug 2026 Official leaderboard
19 Grok 4.6 1,499 ±47.5 Pi Aug 2026 Aug 2026 Official leaderboard
20 Gemini 3.5 Flash 1,496 1477 to 1510 Pi May 2026 Aug 2026 Official leaderboard
21 Gemini 3.1 Pro 1,495 1477 to 1509 Pi Feb 2026 Aug 2026 Official leaderboard
22 Kimi K3 1,495 1454 to 1531 Pi Jul 2026 Aug 2026 Official leaderboard
23 Claude Fable 5.1 1,494 1445 to 1555 Pi Sep 2026 Aug 2026 Official leaderboard
24 Muse Spark 1.3 1,494 1447 to 1535 Pi Sep 2026 Aug 2026 Official leaderboard
25 GLM-5.2 1,492 1466 to 1512 Pi Jun 2026 Aug 2026 Official leaderboard
26 GLM-5.3 1,492 1433 to 1542 Pi Aug 2026 Aug 2026 Official leaderboard
27 Gemini 3.6 Flash 1,489 ±44.5 Pi Jul 2026 Aug 2026 Official leaderboard
28 Grok 4.5 1,487 1451 to 1525 Pi Jul 2026 Aug 2026 Official leaderboard
29 GPT-5.6 Luna 1,486 1460 to 1510 Pi Jul 2026 Aug 2026 Official leaderboard
30 Gemini 3.8 Flash 1,479 1443 to 1520 Pi Sep 2026 Aug 2026 Official leaderboard
31 DeepSeek-V4-Flash 1,476 1453 to 1493 Pi Apr 2026 Aug 2026 Official leaderboard
32 Nemotron 3 Ultra 1,468 1439 to 1495 Pi Jun 2026 Aug 2026 Official leaderboard
33 Gemini 3.7 Flash 1,468 1425 to 1507 Pi Aug 2026 Aug 2026 Official leaderboard
34 Gemini 3.5 Flash-Lite 1,461 1416 to 1501 Pi Jul 2026 Aug 2026 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard Not tested since Sep 2026

Source: Official Multi-Agent Arena Social Poker leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
1,460 1,480 1,500 1,520 1,540 1,560 2026 1,460 1,480 1,500 1,520 1,540 1,560 2026

Source: Official Multi-Agent Arena Social Poker leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Games and long tasks

All 261 in Games and long tasks
  • ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6.2 Astra 100%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games. Gemini 2.5 Pro 3.90
  • MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks. Seed2.0 Pro 49%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress. Llama 4 Scout 7.00
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
  • Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (34 entries).
  • Pass Models scored 34 models we track.
  • Pass Newest score The newest is from September 2026, for GPT-6.1 Sol.
  • Pass Frontier models 5 of the top 10 have a score, led by GPT-6 Astra (#4).
  • Note Headroom Scored as a rating, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on olamlabs.ai: 34 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check