Benchmark Almanac

Buyout Game

GPT-5.5 1975.4 leads since April 2026
Not in model cards

Eight language models play a money-driven elimination game with private transfers and a buyout endgame.

The public repository reports a leaderboard with GPT-5.4 at 1960.6 final wealth among the listed models.

Made by
Lech Mazur
Released
January 2026
Kind
Multi-agent games
Level
Everyday
Scored
Settled by events
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions, Text
Interaction
Multi-agent
Environment
Buyout Game
Items from
Live play or votes
Refresh
Live
Run by
Anyone
In views
Games, Agents, Kept fresh
Strongest model tested
GPT-5.5 #21 on the capability index
Last model added
May 2026
Scores refreshed
November 20th, 2026
32
Models scored on its own leaderboard

GPT-5.5 leads with 1975.4 among models released by May 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. Every run is in the next chart.

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

  1. 1 GPT-5.5 high reasoning 1975.4
  2. 2 GPT-5.4 high reasoning 1960.6
  3. 3 Claude Opus 4.7 high reasoning 1878.4
  4. 4 GLM-5 1791.6
  5. 5 Claude Opus 4.6 high reasoning 1759.7
  6. 6 DeepSeek-V4-Pro 1681.6
  7. 7 Gemini 3.5 Flash 1667.2
  8. 8 Claude Sonnet 4.6 high reasoning 1656.2
  9. 9 Tencent Hy3 preview high 1620.9
  10. 10 Gemini 3.1 Flash-Lite 1615.3
  11. 11 GLM-5.1 1604.9
  12. 12 Gemini 3.1 Pro 1563.8
  13. 13 DeepSeek-V4-Flash 1555.8
  14. 14 Grok 4.20 Reasoning 1549.4
  15. 15 Grok 4.3 1525.1
  16. 16 Gemma 4 31B IT Reasoning 1522.0
  17. 17 MiMo-V2.5-Pro 1516.5
  18. 18 Kimi K2.6 1467.6
  19. 19 Kimi K2.5 Thinking 1461.7
  20. 20 Qwen 3.6 Max (Preview) 1458.7
  21. 21 MiMo-V2-Pro 1427.5
  22. 22 Step 3.5 Flash 1415.5
  23. 23 Seed2.0 Pro 1404.6
  24. 24 DeepSeek-V3.2 1380.7
  25. 25 MiniMax-M2.7 1374.2
  26. 26 Mistral Medium 3.5 high 1332.2
  27. 27 Qwen3.5 397B-A17B 1309.6
  28. 28 ERNIE 5.1 1210.5
  29. 29 Mistral Large 3 1206.3
  30. 30 Llama 4 Maverick 1127.5
  31. 31 ERNIE 5.0 1043.0
  32. 32 gpt-oss-120b 997.1

Source: Official Buyout Game leaderboard

#ModelScore ±SettingDateSource
1 GPT-5.5 1975.4 ±100.4 high reasoning Apr 2026 Official leaderboard
2 GPT-5.4 1960.6 ±71.3 high reasoning Mar 2026 Official leaderboard
3 Claude Opus 4.7 1878.4 ±63.1 high reasoning Apr 2026 Official leaderboard
4 GLM-5 1791.6 ±66 Feb 2026 Official leaderboard
5 Claude Opus 4.6 1759.7 ±65.2 high reasoning Feb 2026 Official leaderboard
6 DeepSeek-V4-Pro 1681.6 ±90.8 Apr 2026 Official leaderboard
7 Gemini 3.5 Flash 1667.2 ±88.4 May 2026 Official leaderboard
8 Claude Sonnet 4.6 1656.2 ±47 high reasoning Feb 2026 Official leaderboard
9 Tencent Hy3 preview 1620.9 ±91.8 high Apr 2026 Official leaderboard
10 Gemini 3.1 Flash-Lite 1615.3 ±54.7 Mar 2026 Official leaderboard
11 GLM-5.1 1604.9 ±89.2 Apr 2026 Official leaderboard
12 Gemini 3.1 Pro 1563.8 ±46.9 Feb 2026 Official leaderboard
13 DeepSeek-V4-Flash 1555.8 ±95.3 Apr 2026 Official leaderboard
14 Grok 4.20 1549.4 ±63.9 Reasoning Feb 2026 Official leaderboard
15 Grok 4.3 1525.1 ±92.6 Apr 2026 Official leaderboard
16 Gemma 4 31B IT 1522.0 ±95.5 Reasoning Apr 2026 Official leaderboard
17 MiMo-V2.5-Pro 1516.5 ±99.1 Apr 2026 Official leaderboard
18 Kimi K2.6 1467.6 ±90.8 Apr 2026 Official leaderboard
19 Kimi K2.5 1461.7 ±59.2 Thinking Jan 2026 Official leaderboard
20 Qwen 3.6 Max (Preview) 1458.7 ±182.2 Apr 2026 Official leaderboard
21 MiMo-V2-Pro 1427.5 ±59.1 Mar 2026 Official leaderboard
22 Step 3.5 Flash 1415.5 ±88.4 Feb 2026 Official leaderboard
23 Seed2.0 Pro 1404.6 ±46.2 Feb 2026 Official leaderboard
24 DeepSeek-V3.2 1380.7 ±55.8 Dec 2025 Official leaderboard
25 MiniMax-M2.7 1374.2 ±45.8 Mar 2026 Official leaderboard
26 Mistral Medium 3.5 1332.2 ±89.2 high Apr 2026 Official leaderboard
27 Qwen3.5 397B-A17B 1309.6 ±52.8 Feb 2026 Official leaderboard
28 ERNIE 5.1 1210.5 ±98.3 May 2026 Official leaderboard
29 Mistral Large 3 1206.3 ±66.2 Dec 2025 Official leaderboard
30 Llama 4 Maverick 1127.5 ±67.5 Apr 2025 Official leaderboard
31 ERNIE 5.0 1043.0 ±58.2 Jan 2026 Official leaderboard
32 gpt-oss-120b 997.1 ±51.3 Aug 2025 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-5.4
no reasoning high reason…
1960.6
Claude Opus 4.6
no reasoning high reason…
1759.7
Claude Sonnet 4.6
no reasoning high reason…
1656.2
Grok 4.20
Non-Reasoni… Reasoning
1549.4

Source: Official Buyout Game leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.5 high reasoning 1975.4
GPT-5.4 high reasoning 1960.6
Claude Opus 4.7 high reasoning 1878.4
GLM-5 1791.6
Claude Opus 4.6 high reasoning 1759.7
DeepSeek-V4-Pro 1681.6
Gemini 3.5 Flash 1667.2
Claude Sonnet 4.6 high reasoning 1656.2
GPT-5.4 no reasoning 1654.0
Tencent Hy3 preview high 1620.9
Gemini 3.1 Flash-Lite 1615.3
GLM-5.1 1604.9
Gemini 3.1 Pro 1563.8
DeepSeek-V4-Flash 1555.8
Claude Sonnet 4.6 no reasoning 1549.9
Grok 4.20 Reasoning 1549.4
Grok 4.3 1525.1
Gemma 4 31B IT Reasoning 1522.0
MiMo-V2.5-Pro 1516.5
Claude Opus 4.6 no reasoning 1510.2
Kimi K2.6 1467.6
Kimi K2.5 Thinking 1461.7
Qwen 3.6 Max (Preview) 1458.7
MiMo-V2-Pro 1427.5
Step 3.5 Flash 1415.5
Seed2.0 Pro 1404.6
DeepSeek-V3.2 1380.7
MiniMax-M2.7 1374.2
Mistral Medium 3.5 high 1332.2
Qwen3.5 397B-A17B 1309.6
Grok 4.20 Non-Reasoning 1224.7
ERNIE 5.1 1210.5
Mistral Large 3 1206.3
Llama 4 Maverick 1127.5
ERNIE 5.0 1043.0
gpt-oss-120b 997.1

6 models have held number one on the official leaderboard since its release in January 2026

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

DeepSeek Moonshot AI Anthropic Z.ai OpenAI
Own leaderboard Not tested since May 2026
GPT-5.5 1975.4

Source: Official Buyout Game 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
1000.0 1250.0 1500.0 1750.0 2000.0 2025 2026 1000.0 1250.0 1500.0 1750.0 2000.0 2025 2026

Source: Official Buyout Game 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 6 months ago, in May 2026, for Gemini 3.5 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 (32 entries).
  • Pass Models scored 32 models we track.
  • Warning Newest score The newest came 6 months ago, in May 2026, for Gemini 3.5 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.5 (#21).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 36 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