LLM Public Goods Game
Agents choose how many tokens to contribute to a common pool. The evaluation reports the fraction each model contributes across repeated games.
The leaderboard compares 21 models over 310 games. The later PGG-Bench also allows punishment.
- Made by
- Lech Mazur
- Released
- February 2025
- Kind
- Multi-agent games
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Propensity
- Behavior
- Misaligned with values
- Language
- English
- Interaction
- Multi-agent
- Environment
- Public goods game
- Items from
- Live play or votes
- Refresh
- Live
- Run by
- Anyone
- In views
- Games, Agents, Kept fresh
- Strongest model tested
- o1 #118 on the capability index
- Last model added
- May 2025
- Scores refreshed
- November 21st, 2026
Gemini 2.0 Flash leads with 45.2% among models released by May 2025 Gemini 2.0 Flash leads with 45.2% among models released by May 2025
Best score per entry from the benchmark's own leaderboard, read on November 21st, 2026. Where a model ran with several settings or harnesses, we keep its best. 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 May 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 45.2
Gemini 2.0 Flash
- 41.0 Claude 3.5 Haiku
- 31.2
Gemma 2 27B
- 30.2
Gemini 1.5 Flash
- 24.8 GPT-4o mini
- 24.6 Gemini 1.5 Pro (Sept)
- 23.6
Qwen2.5-72B
- 20.9 Claude 3.5 Sonnet
- 20.0
Gemini 2.0 Flash Thinking
- 15.8 Grok-2
- 15.5
DeepSeek-V3
- 14.7 GPT-4o
- 3.16
Llama 3.1-405B
- 2.59 o1-mini
- 2.55 o1
- 1.57
Mistral Large 2
- 1.24
Llama 3.3 70B
- 0.51
Qwen2.5-Max
- 0.31
DeepSeek-R1
- 0.0 o3-mini
- 1
Gemini 2.0 Flash Feb 2025 45.2%
- 2 Claude 3.5 Haiku 41%
- 3
Gemma 2 27B 31.2%
- 4
Gemini 1.5 Flash Sep 2024 30.2%
- 5 GPT-4o mini 24.8%
- 6 Gemini 1.5 Pro (Sept) 24.6%
- 7
Qwen2.5-72B 23.6%
- 8 Claude 3.5 Sonnet 20.9%
- 9
Gemini 2.0 Flash Thinking Jan 2025 20%
- 10 Grok-2 Dec 2024 15.8%
- 11
DeepSeek-V3 Mar 2025 15.5%
- 12 GPT-4o Mar 2025 14.7%
- 13
Llama 3.1-405B 3.16%
- 14 o1-mini 2.59%
- 15 o1 2.55%
- 16
Mistral Large 2 Nov 2024 1.57%
- 17
Llama 3.3 70B 1.24%
- 18
Qwen2.5-Max 0.51%
- 19
DeepSeek-R1 May 2025 0.31%
- 20 o3-mini 0%
Source: Official LLM Public Goods Game leaderboard
- 45.2
Gemini 2.0 Flash
- 41.0 Claude 3.5 Haiku
- 31.2
Gemma 2 27B
- 30.2
Gemini 1.5 Flash
- 24.8 GPT-4o mini
- 23.6
Qwen2.5-72B
- 20.9 Claude 3.5 Sonnet
- 20.0
Gemini 2.0 Flash Thinking
- 15.8 Grok-2
- 15.5
DeepSeek-V3
- 14.7 GPT-4o
- 3.16
Llama 3.1-405B
- 2.59 o1-mini
- 2.55 o1
- 1.57
Mistral Large 2
- 1.24
Llama 3.3 70B
- 0.51
Qwen2.5-Max
- 0.31
DeepSeek-R1
- 0.0 o3-mini
- 1
Gemini 2.0 Flash Feb 2025 45.2%
- 2 Claude 3.5 Haiku 41%
- 3
Gemma 2 27B 31.2%
- 4
Gemini 1.5 Flash Sep 2024 30.2%
- 5 GPT-4o mini 24.8%
- 6
Qwen2.5-72B 23.6%
- 7 Claude 3.5 Sonnet 20.9%
- 8
Gemini 2.0 Flash Thinking Jan 2025 20%
- 9 Grok-2 Dec 2024 15.8%
- 10
DeepSeek-V3 Mar 2025 15.5%
- 11 GPT-4o Mar 2025 14.7%
- 12
Llama 3.1-405B 3.16%
- 13 o1-mini 2.59%
- 14 o1 2.55%
- 15
Mistral Large 2 Nov 2024 1.57%
- 16
Llama 3.3 70B 1.24%
- 17
Qwen2.5-Max 0.51%
- 18
DeepSeek-R1 May 2025 0.31%
- 19 o3-mini 0%
Source: Official LLM Public Goods Game leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 45.2% | Feb 2025 | Official leaderboard | |
| 2 | Claude 3.5 Haiku | 41% | Oct 2024 | Official leaderboard | |
| 3 | | 31.2% | Jun 2024 | Official leaderboard | |
| 4 | | 30.2% | Aug 2024 | Official leaderboard | |
| 5 | GPT-4o mini | 24.8% | Jul 2024 | Official leaderboard | |
| 6 | Gemini 1.5 Pro (Sept) | 24.6% | Official leaderboard | ||
| 7 | | 23.6% | Sep 2024 | Official leaderboard | |
| 8 | Claude 3.5 Sonnet | 20.9% | Jun 2024 | Official leaderboard | |
| 9 | | 20% | Jan 2025 | Official leaderboard | |
| 10 | Grok-2 Dec 2024 | 15.8% | Dec 2024 | Official leaderboard | |
| 11 | | 15.5% | Mar 2025 | Official leaderboard | |
| 12 | GPT-4o Mar 2025 | 14.7% | Mar 2025 | Official leaderboard | |
| 13 | | 3.16% | Jul 2024 | Official leaderboard | |
| 14 | o1-mini | 2.59% | Sep 2024 | Official leaderboard | |
| 15 | o1 | 2.55% | Dec 2024 | Official leaderboard | |
| 16 | | 1.57% | Nov 2024 | Official leaderboard | |
| 17 | | 1.24% | Dec 2024 | Official leaderboard | |
| 18 | | 0.51% | Jan 2025 | Official leaderboard | |
| 19 | | 0.31% | May 2025 | Official leaderboard | |
| 20 | o3-mini | 0% | Jan 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 45.2% | Feb 2025 | Official leaderboard | |
| 2 | Claude 3.5 Haiku | 41% | Oct 2024 | Official leaderboard | |
| 3 | | 31.2% | Jun 2024 | Official leaderboard | |
| 4 | | 30.2% | Aug 2024 | Official leaderboard | |
| 5 | GPT-4o mini | 24.8% | Jul 2024 | Official leaderboard | |
| 6 | | 23.6% | Sep 2024 | Official leaderboard | |
| 7 | Claude 3.5 Sonnet | 20.9% | Jun 2024 | Official leaderboard | |
| 8 | | 20% | Jan 2025 | Official leaderboard | |
| 9 | Grok-2 Dec 2024 | 15.8% | Dec 2024 | Official leaderboard | |
| 10 | | 15.5% | Mar 2025 | Official leaderboard | |
| 11 | GPT-4o Mar 2025 | 14.7% | Mar 2025 | Official leaderboard | |
| 12 | | 3.16% | Jul 2024 | Official leaderboard | |
| 13 | o1-mini | 2.59% | Sep 2024 | Official leaderboard | |
| 14 | o1 | 2.55% | Dec 2024 | Official leaderboard | |
| 15 | | 1.57% | Nov 2024 | Official leaderboard | |
| 16 | | 1.24% | Dec 2024 | Official leaderboard | |
| 17 | | 0.51% | Jan 2025 | Official leaderboard | |
| 18 | | 0.31% | May 2025 | Official leaderboard | |
| 19 | o3-mini | 0% | Jan 2025 | Official leaderboard |
2 models have held number one on the official leaderboard since its release in February 2025
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official LLM Public Goods Game leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official LLM Public Goods 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
- 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
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 18 months ago, in May 2025, for DeepSeek-R1. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the official leaderboard (20 entries).
- Pass Models scored 19 models we track, plus 1 other entry on its leaderboard.
- Warning Newest score The newest came 18 months ago, in May 2025, for DeepSeek-R1.
- Note Frontier models None of the top 10 has a score. The strongest tested is o1 (#118).
- Pass Headroom The best is 45.2%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on github.com: 20 rows for models we track and 1 other entry at the last read.
- Pass Last read November 21st, 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.