Benchmark Almanac

Peer Arena (oddbit.ai)

GPT-5.1 1,799 leads since November 2025
Not in model cards

Five models debate for five rounds, then each privately votes for a survivor. The organizer calculates Bradley-Terry ratings from game outcomes.

The maker ranks 17 models in both experiments. Participants saw model names in 198 games and anonymous labels in 100 games.

Made by
Can Celik
Kind
Multi-agent games
Level
Everyday
Scored
Ranked by matches
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Multi-agent
Environment
Peer Arena
Items from
Live play or votes
Refresh
Live
Run by
Anyone
In views
Games, Agents, Kept fresh
Strongest model tested
GPT-5.2 #51 on the capability index
Last model added
December 2025
Scores refreshed
November 20th, 2026
17
Entries on its own leaderboard, 16 models we track

GPT-5.1 leads with 1,799 among models released by December 2025 GPT-5.1 leads with 1,799 among models released by December 2025

Best score per entry 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. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in December 2025, 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 1,000.

  1. 1 GPT-5.1 anonymous 1,799
  2. 2 Claude Opus 4.5 identity 1,699
  3. 3 GPT-5.2 identity 1,687
  4. 4 Claude Haiku 4.5 anonymous 1,611
  5. 5 Claude Sonnet 4.5 identity 1,599
  6. 6 gpt-oss-120b identity 1,533
  7. 7 Kimi K2 Thinking identity 1,520
  8. 8 Qwen3-Max anonymous 1,453
  9. 9 Gemini 3 Pro identity 1,387
  10. 10 Claude 3 Opus identity 1,370
  11. 11 MiniMax-M2 anonymous 1,364
  12. 12 Gemini 3 Flash identity 1,358
  13. 13 mistral-large-2512 identity 1,350
  14. 14 DeepSeek-V3.2 identity 1,306
  15. 15 Grok 4 anonymous 1,217
  16. 16 GLM-4.6 identity 1,041
  17. 17 GPT-4o Mar 2025, identity 1,016

Source: Official Peer Arena (oddbit.ai) leaderboard

#EntryScoreSettingDateLinkSource
1 GPT-5.1 1,799 anonymous Nov 2025 oddbit.ai Official leaderboard
2 Claude Opus 4.5 1,699 identity Nov 2025 oddbit.ai Official leaderboard
3 GPT-5.2 1,687 identity Dec 2025 oddbit.ai Official leaderboard
4 Claude Haiku 4.5 1,611 anonymous Oct 2025 oddbit.ai Official leaderboard
5 Claude Sonnet 4.5 1,599 identity Sep 2025 oddbit.ai Official leaderboard
6 gpt-oss-120b 1,533 identity Aug 2025 oddbit.ai Official leaderboard
7 Kimi K2 Thinking 1,520 identity Nov 2025 oddbit.ai Official leaderboard
8 Qwen3-Max 1,453 anonymous Sep 2025 oddbit.ai Official leaderboard
9 Gemini 3 Pro 1,387 identity Nov 2025 oddbit.ai Official leaderboard
10 Claude 3 Opus 1,370 identity Feb 2024 oddbit.ai Official leaderboard
11 MiniMax-M2 1,364 anonymous Oct 2025 oddbit.ai Official leaderboard
12 Gemini 3 Flash 1,358 identity Dec 2025 oddbit.ai Official leaderboard
13 mistral-large-2512 1,350 identity oddbit.ai Official leaderboard
14 DeepSeek-V3.2 1,306 identity Dec 2025 oddbit.ai Official leaderboard
15 Grok 4 1,217 anonymous Jul 2025 oddbit.ai Official leaderboard
16 GLM-4.6 1,041 identity Sep 2025 oddbit.ai Official leaderboard
17 GPT-4o Mar 2025 1,016 identity Mar 2025 oddbit.ai Official leaderboard
#ModelScoreSettingDateLinkSource
1 GPT-5.1 1,799 anonymous Nov 2025 oddbit.ai Official leaderboard
2 Claude Opus 4.5 1,699 identity Nov 2025 oddbit.ai Official leaderboard
3 GPT-5.2 1,687 identity Dec 2025 oddbit.ai Official leaderboard
4 Claude Haiku 4.5 1,611 anonymous Oct 2025 oddbit.ai Official leaderboard
5 Claude Sonnet 4.5 1,599 identity Sep 2025 oddbit.ai Official leaderboard
6 gpt-oss-120b 1,533 identity Aug 2025 oddbit.ai Official leaderboard
7 Kimi K2 Thinking 1,520 identity Nov 2025 oddbit.ai Official leaderboard
8 Qwen3-Max 1,453 anonymous Sep 2025 oddbit.ai Official leaderboard
9 Gemini 3 Pro 1,387 identity Nov 2025 oddbit.ai Official leaderboard
10 Claude 3 Opus 1,370 identity Feb 2024 oddbit.ai Official leaderboard
11 MiniMax-M2 1,364 anonymous Oct 2025 oddbit.ai Official leaderboard
12 Gemini 3 Flash 1,358 identity Dec 2025 oddbit.ai Official leaderboard
13 DeepSeek-V3.2 1,306 identity Dec 2025 oddbit.ai Official leaderboard
14 Grok 4 1,217 anonymous Jul 2025 oddbit.ai Official leaderboard
15 GLM-4.6 1,041 identity Sep 2025 oddbit.ai Official leaderboard
16 GPT-4o Mar 2025 1,016 identity Mar 2025 oddbit.ai Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-5.1
identity anonymous
1,799
Claude Opus 4.5
anonymous identity
1,699
GPT-5.2
anonymous
1,687
Claude Haiku 4.5
identity anonymous
1,611
Claude Sonnet 4.5
anonymous identity
1,599
gpt-oss-120b
anonymous identity
1,533
Kimi K2 Thinking
anonymous identity
1,520
Qwen3-Max
identity anonymous
1,453
Gemini 3 Pro
anonymous identity
1,387
Claude 3 Opus
anonymous identity
1,370
MiniMax-M2
identity anonymous
1,364
Gemini 3 Flash
anonymous
1,358

Source: Official Peer Arena (oddbit.ai) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.1 anonymous 1,799
Claude Opus 4.5 identity 1,699
GPT-5.1 identity 1,691
GPT-5.2 identity 1,687
GPT-5.2 anonymous 1,665
Claude Opus 4.5 anonymous 1,639
Claude Haiku 4.5 anonymous 1,611
Claude Sonnet 4.5 identity 1,599
gpt-oss-120b identity 1,533
Kimi K2 Thinking identity 1,520
Claude Sonnet 4.5 anonymous 1,510
Qwen3-Max anonymous 1,453
Claude Haiku 4.5 identity 1,439
Gemini 3 Pro identity 1,387
Qwen3-Max identity 1,375
Claude 3 Opus identity 1,370
Kimi K2 Thinking anonymous 1,369
MiniMax-M2 anonymous 1,364
Gemini 3 Flash identity 1,358
Gemini 3 Flash anonymous 1,334
DeepSeek-V3.2 identity 1,306
Gemini 3 Pro anonymous 1,289
gpt-oss-120b anonymous 1,272
DeepSeek-V3.2 anonymous 1,240
MiniMax-M2 identity 1,219
Grok 4 anonymous 1,217
Claude 3 Opus anonymous 1,200
Grok 4 identity 1,121
GLM-4.6 identity 1,041
GPT-4o Mar 2025 identity 1,016
GLM-4.6 anonymous 955
GPT-4o Mar 2025 anonymous 814

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-5.1
identity anonymous
1,799
Claude Opus 4.5
anonymous identity
1,699
GPT-5.2
anonymous
1,687
Claude Haiku 4.5
identity anonymous
1,611
Claude Sonnet 4.5
anonymous identity
1,599
gpt-oss-120b
anonymous identity
1,533
Kimi K2 Thinking
anonymous identity
1,520
Qwen3-Max
identity anonymous
1,453
Gemini 3 Pro
anonymous identity
1,387
Claude 3 Opus
anonymous identity
1,370
MiniMax-M2
identity anonymous
1,364
Gemini 3 Flash
anonymous
1,358

Source: Official Peer Arena (oddbit.ai) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.1 anonymous 1,799
Claude Opus 4.5 identity 1,699
GPT-5.1 identity 1,691
GPT-5.2 identity 1,687
GPT-5.2 anonymous 1,665
Claude Opus 4.5 anonymous 1,639
Claude Haiku 4.5 anonymous 1,611
Claude Sonnet 4.5 identity 1,599
gpt-oss-120b identity 1,533
Kimi K2 Thinking identity 1,520
Claude Sonnet 4.5 anonymous 1,510
Qwen3-Max anonymous 1,453
Claude Haiku 4.5 identity 1,439
Gemini 3 Pro identity 1,387
Qwen3-Max identity 1,375
Claude 3 Opus identity 1,370
Kimi K2 Thinking anonymous 1,369
MiniMax-M2 anonymous 1,364
Gemini 3 Flash identity 1,358
Gemini 3 Flash anonymous 1,334
DeepSeek-V3.2 identity 1,306
Gemini 3 Pro anonymous 1,289
gpt-oss-120b anonymous 1,272
DeepSeek-V3.2 anonymous 1,240
MiniMax-M2 identity 1,219
Grok 4 anonymous 1,217
Claude 3 Opus anonymous 1,200
Grok 4 identity 1,121
GLM-4.6 identity 1,041
GPT-4o Mar 2025 identity 1,016
GLM-4.6 anonymous 955
GPT-4o Mar 2025 anonymous 814

5 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.

Anthropic OpenAI
Own leaderboard Not tested since Dec 2025
GPT-5.1 1,799

Source: Official Peer Arena (oddbit.ai) 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
800 1,000 1,200 1,400 1,600 1,800 2024 2025 2026 800 1,000 1,200 1,400 1,600 1,800 2024 2025 2026

Source: Official Peer Arena (oddbit.ai) 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

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 11 months ago, in December 2025, for Gemini 3 Flash. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (17 entries).
  • Pass Models scored 16 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 11 months ago, in December 2025, for Gemini 3 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
  • Note Headroom Scored as a rating, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on oddbit.ai: 32 rows for models we track and 1 other entry 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