ChessBench.org - Glicko-2 Rating - Protocol v3: leaderboard
ChessBench.org's saved chessbench-v3 qualified setup ratings in an LLM-only Glicko-2 pool. Every native model/effort/setup/profile identity stays distinct; ratings are unbounded and are not engine-calibrated or human chess ratings. Source-qualified and provisional flags, rated game counts and W/D/L remain explicit. The current methodology says prompt v3.2; it is not assigned retroactively to this older aggregate's source-defined prompt population.
Metric: Glicko-2 chess rating (unbounded; LLM-only pool). Source: chessbench.org. Saturation forecast: Rough model projection: around 2026. 32 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon (ChessBench.org Run: protocol=chessbench-v3;effort=medium;setup=975229c8be30534961de;profile=a268b289-192f-4126-9c0f-3ee4932fe2bb) | 2170.37 | |
| 2 | Claude Opus 5.5 (ChessBench.org Run: protocol=chessbench-v3;effort=xhigh;setup=fc64f6c9781b8572d810;profile=361225c1-c813-48e2-aa2c-f5cb3d2721d5) | 2102.12 | |
| 3 | Claude Sonnet 5.5 (ChessBench.org Run: protocol=chessbench-v3;effort=xhigh;setup=a3f0893780fecfa9d830;profile=cbebb32a-7dae-4c1f-8bb0-59d84361b16e) | 2089.21 | |
| 4 | Gemini 4 Argon (ChessBench.org Run: protocol=chessbench-v3;effort=high;setup=61809341af9571906acd;profile=db04eac7-af35-42a2-8d5c-e0be46d70b36) | 2087.52 | |
| 5 | GPT-6 Astra (ChessBench.org Run: protocol=chessbench-v3;effort=xhigh;setup=f2b1a7b93f215cb6f16f;profile=358b1292-08f6-4078-b2ee-63936e2e5711) | 2022.45 | |
| 6 | Claude Opus 5.5 (ChessBench.org Run: protocol=chessbench-v3;effort=medium;setup=ca1679a6ee8397405382;profile=3715708e-c45c-4272-a9ad-18730ea69f8b) | 1996.97 | |
| 7 | Claude Fable 5.1 (ChessBench.org Run: protocol=chessbench-v3;effort=xhigh;setup=d0db2b308232ac4f749a;profile=2677f832-ac32-454f-8473-d68056dd2431) | 1890.71 | |
| 8 | GPT-6.1 Sol (ChessBench.org Run: protocol=chessbench-v3;effort=xhigh;setup=6171b0456e3b7841d58e;profile=5a037695-46d8-4827-ad93-5a216ce9bc0b) | 1840.03 | |
| 9 | Claude Haiku 5.5 (ChessBench.org Run: protocol=chessbench-v3;effort=xhigh;setup=07323894ce073259258d;profile=31accff9-5d60-4ceb-87fc-bbbab156dc24) | 1796.08 | |
| 10 | GPT-6 Astra (ChessBench.org Run: protocol=chessbench-v3;effort=medium;setup=bc5b15335b6158973748;profile=6ab660a2-a8a9-42cd-8f3a-c57e212f5b9c) | 1767.17 |
Interactive version: theaggregate.ai/benchmark?slug=chessbench-org-glicko-2-rating-protocol-v3 · How It Works · Data refreshed daily, snapshot 2026-11-05.