Benchmark Almanac

ChessBench (MadeByLupu)

Gemini 4 Argon 2,170 leads since September 2026
Not in model cards

Models play each other with access to the board, move history and legal moves. The organizer rates each model and thinking setting with Glicko-2.

The standings include 18 distinct models. Placement pairs play both colors, and the organizer counts unfinished games at the move limit as draws.

Made by
MadeByLupu
Kind
Multi-agent games
Level
School
Scored
Ranked by matches
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Multi-agent
Environment
Chess
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
October 2026
Scores refreshed
November 20th, 2026
18
Models scored on its own leaderboard

Gemini 4 Argon leads with 2,170

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. Every run is in the next chart.

Bars start at 1,000.

  1. 1 Gemini 4 Argon medium 2,170
  2. 2 Claude Opus 5.5 xhigh 2,102
  3. 3 Claude Sonnet 5.5 xhigh 2,089
  4. 4 GPT-6 Astra xhigh 2,022
  5. 5 Claude Fable 5.1 xhigh 1,891
  6. 6 GPT-6.1 Sol xhigh 1,840
  7. 7 Claude Haiku 5.5 xhigh 1,796
  8. 8 DeepSeek V4.1 Flash 1017 max 1,687
  9. 9 GPT-6.1 Terra xhigh 1,621
  10. 10 Grok 4.7 xhigh 1,549
  11. 11 Muse Spark 1.3 medium 1,506
  12. 12 GLM-5.3 high 1,501
  13. 13 GPT-6 Luna xhigh 1,421
  14. 14 MiniMax-M3.1-Flash xhigh 1,386
  15. 15 MiMo-V2.6 Flash enabled 1,345
  16. 16 GLM-5.3-Flash high 1,327
  17. 17 MiniMax-M3 adaptive 1,261
  18. 18 Qwen3.7-Plus enabled 1,075

Source: Official ChessBench (MadeByLupu) leaderboard

#ModelScoreSettingReleasedRun onLinkSource
1 Gemini 4 Argon 2,170 medium Sep 2026 Oct 2026 chessbench.org Official leaderboard
2 Claude Opus 5.5 2,102 xhigh Sep 2026 Oct 2026 chessbench.org Official leaderboard
3 Claude Sonnet 5.5 2,089 xhigh Sep 2026 Oct 2026 chessbench.org Official leaderboard
4 GPT-6 Astra 2,022 xhigh Sep 2026 Oct 2026 chessbench.org Official leaderboard
5 Claude Fable 5.1 1,891 xhigh Sep 2026 Oct 2026 chessbench.org Official leaderboard
6 GPT-6.1 Sol 1,840 xhigh Sep 2026 Oct 2026 chessbench.org Official leaderboard
7 Claude Haiku 5.5 1,796 xhigh Oct 2026 Oct 2026 chessbench.org Official leaderboard
8 DeepSeek V4.1 Flash 1017 1,687 max Oct 2026 Oct 2026 chessbench.org Official leaderboard
9 GPT-6.1 Terra 1,621 xhigh Oct 2026 Oct 2026 chessbench.org Official leaderboard
10 Grok 4.7 1,549 xhigh Sep 2026 Oct 2026 chessbench.org Official leaderboard
11 Muse Spark 1.3 1,506 medium Sep 2026 Oct 2026 chessbench.org Official leaderboard
12 GLM-5.3 1,501 high Aug 2026 Oct 2026 chessbench.org Official leaderboard
13 GPT-6 Luna 1,421 xhigh Sep 2026 Oct 2026 chessbench.org Official leaderboard
14 MiniMax-M3.1-Flash 1,386 xhigh Oct 2026 Oct 2026 chessbench.org Official leaderboard
15 MiMo-V2.6 Flash 1,345 enabled Sep 2026 Oct 2026 chessbench.org Official leaderboard
16 GLM-5.3-Flash 1,327 high Aug 2026 Oct 2026 chessbench.org Official leaderboard
17 MiniMax-M3 1,261 adaptive Jun 2026 Oct 2026 chessbench.org Official leaderboard
18 Qwen3.7-Plus 1,075 enabled Jun 2026 Oct 2026 chessbench.org Official leaderboard

The best score, 2,170, costs $0.41 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Meta OpenAI Google DeepMind Anthropic DeepSeek Cost frontier One model at each effort level
0 500 1,000 1,500 2,000 2,500 $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Gemini 4 Argon, medium GPT-6.1 Sol, xhigh DeepSeek V4.1 Flash 1017, max Claude Haiku 5.5, medium GPT-6 Luna, medium 0 500 1,000 1,500 2,000 2,500 $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Gemini 4 Argon, medium GPT-6.1 Sol, xhigh

Source: Official ChessBench (MadeByLupu) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Argon medium 2,170 $0.41 11,855
Claude Opus 5.5 xhigh 2,102 $1.78 42,700
Claude Sonnet 5.5 xhigh 2,089 $2.23 146,849
Gemini 4 Argon high 2,088 $0.72 33,171
GPT-6 Astra xhigh 2,022 $1.45 9,068
Claude Opus 5.5 medium 1,997 $0.67 10,203
Claude Fable 5.1 xhigh 1,891 $8.31 97,530
GPT-6.1 Sol xhigh 1,840 $0.36 16,499
Claude Haiku 5.5 xhigh 1,796 $0.28 118,965
GPT-6 Astra medium 1,767 $1.03 1,798
DeepSeek V4.1 Flash 1017 max 1,687 $0.20 217,586
Claude Sonnet 5.5 medium 1,641 $0.45 14,546
GPT-6.1 Sol medium 1,630 $0.17 2,473
GPT-6.1 Terra xhigh 1,621 $0.11 5,479
DeepSeek V4.1 Flash 1017 high 1,612 $0.21 214,279
Claude Fable 5.1 medium 1,602 $2.22 16,341
Claude Haiku 5.5 medium 1,584 $0.04 32,609
Grok 4.7 xhigh 1,549 $3.62 151,870
Muse Spark 1.3 medium 1,506 $0.03 66,361
GLM-5.3 high 1,501 $2.13 157,253
GLM-5.3 max 1,491 $1.86 138,162
GPT-6.1 Terra medium 1,451 $0.08 893
Muse Spark 1.3 xhigh 1,426 $0.06 177,320
GPT-6 Luna xhigh 1,421 $0.05 57,303
MiniMax-M3.1-Flash xhigh 1,386 $0.27 176,046
MiMo-V2.6 Flash enabled 1,345 $0.06 105,426
GLM-5.3-Flash high 1,327 $0.11 107,500
GLM-5.3-Flash max 1,279 $0.10 109,400
MiniMax-M3 adaptive 1,261 $0.54 165,895
MiniMax-M3.1-Flash medium 1,189 $0.12 67,657
Qwen3.7-Plus enabled 1,075 $0.07 13,454
GPT-6 Luna medium 932 $0.007 795

6 models have held number one on the official leaderboard since 2023

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

MiniMax Z.ai Anthropic OpenAI Google DeepMind
Own leaderboard 6 leaders

Source: Official ChessBench (MadeByLupu) 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
1,000 1,500 2,000 2026 2027 1,000 1,500 2,000 2026 2027

Source: Official ChessBench (MadeByLupu) 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 All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (18 entries).
  • Pass Models scored 18 models we track.
  • Pass Newest score The newest is from October 2026, for GPT-6.1 Terra.
  • Pass Frontier models 7 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 chessbench.org: 32 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