# Multi-agent coordination on Overcooked-AI Mean of CR/AA/Ring/FC/CC layouts Source: [Multi-agent coordination on Overcooked-AI Mean of CR/AA/Ring/FC/CC layouts]() Collection index: [llms.txt]() Data updated: 2026-06-09T07:58:58.198Z Evaluation of multi-agent coordination performance across various Overcooked-AI layouts (Coordination Ring, Asymmetric Advantages, Forced Coordination, etc.), focusing on behavioral metrics like self-play performance, cross-play performance, and communication efficiency. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Multi-agent coordination | | Dataset (including any stated split) | Overcooked-AI Mean of CR/AA/Ring/FC/CC layouts | | Metric | Self-Play Score | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 16 of 16 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Self-Play Score](): Higher is better; unit: SCALAR. - [Cross-Play Score](): Higher is better; unit: SCALAR. - [Performance Gap](): Lower is better; unit: SCALAR. - [Collisions](): Lower is better; unit: SCALAR. - [Handling Frequency](): Higher is better; unit: SCALAR. - [Idle Percentage](): Lower is better; unit: PERCENT. - [Post-hoc KL Divergence](): Lower is better; unit: SCALAR. - [Tokens per Episode](): Lower is better; unit: SCALAR. - [Messages per Round](): Higher is better; unit: SCALAR. ## Recorded results | Method | Self-Play Score | Result date | Report paper | | --- | --- | --- | --- | | GPT-4-Turbo | 182.7 | Not recorded | 6a279e3d90966ade919bf849 | | GPT-o3 | 156.3 | Not recorded | 6a279e3d90966ade919bf849 | | Debate+Judge (GPT-4) | 125.7 | Not recorded | 6a279e3d90966ade919bf849 | | DICE-FT | 121.6 | Not recorded | 6a279e3d90966ade919bf849 | | MA-Debate (235B) | 117.4 | Not recorded | 6a279e3d90966ade919bf849 | | DICE-FT | 114.4 | Not recorded | 6a279e3d90966ade919bf849 | | rStar | 112.3 | Not recorded | 6a279e3d90966ade919bf849 | | DICE-PC | 110.4 | Not recorded | 6a279e3d90966ade919bf849 | | MA-FT (32B) | 109.7 | Not recorded | 6a279e3d90966ade919bf849 | | Qwen3-235B-Inst | 108.2 | Not recorded | 6a279e3d90966ade919bf849 | | MA-Debate (32B) | 106.8 | Not recorded | 6a279e3d90966ade919bf849 | | CAC (learned) | 105.2 | Not recorded | 6a279e3d90966ade919bf849 | | CAC (uniform) | 103.7 | Not recorded | 6a279e3d90966ade919bf849 | | MAD | 101.3 | Not recorded | 6a279e3d90966ade919bf849 | | Role-Play | 98.4 | Not recorded | 6a279e3d90966ade919bf849 | | Qwen3-32B-Inst | 95.7 | Not recorded | 6a279e3d90966ade919bf849 | ### GPT-4-Turbo Self-Play Score: 182.7. Result ID: 6a27c7aefbc373592c2458b1. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-o3 Self-Play Score: 156.3. Result ID: 6a27c7aefbc373592c2458b0. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Debate+Judge (GPT-4) Self-Play Score: 125.7. Result ID: 6a27c7aefbc373592c2458b7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### DICE-FT Self-Play Score: 121.6. Result ID: 6a27c7aefbc373592c2458bd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3×32B"}. This record does not assert independent reproduction. ### MA-Debate (235B) Self-Play Score: 117.4. Result ID: 6a27c7aefbc373592c2458b8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3×235B"}. This record does not assert independent reproduction. ### DICE-FT Self-Play Score: 114.4. Result ID: 6a27c7aefbc373592c2458bc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3×8B"}. This record does not assert independent reproduction. ### rStar Self-Play Score: 112.3. Result ID: 6a27c7aefbc373592c2458b9. [Reporting paper]() Recorded evaluation configuration: {"Size":"3×32B"}. This record does not assert independent reproduction. ### DICE-PC Self-Play Score: 110.4. Result ID: 6a27c7aefbc373592c2458bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3×8B"}. This record does not assert independent reproduction. ### MA-FT (32B) Self-Play Score: 109.7. Result ID: 6a27c7aefbc373592c2458ba. [Reporting paper]() Recorded evaluation configuration: {"Size":"3×32B"}. This record does not assert independent reproduction. ### Qwen3-235B-Inst Self-Play Score: 108.2. Result ID: 6a27c7aefbc373592c2458af. [Reporting paper]() Recorded evaluation configuration: {"Size":"1×235B"}. This record does not assert independent reproduction. ### MA-Debate (32B) Self-Play Score: 106.8. Result ID: 6a27c7aefbc373592c2458b6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3×32B"}. This record does not assert independent reproduction. ### CAC (learned) Self-Play Score: 105.2. Result ID: 6a27c7aefbc373592c2458b5. [Reporting paper]() Recorded evaluation configuration: {"Size":"3×8B"}. This record does not assert independent reproduction. ### CAC (uniform) Self-Play Score: 103.7. Result ID: 6a27c7aefbc373592c2458b4. [Reporting paper]() Recorded evaluation configuration: {"Size":"3×8B"}. This record does not assert independent reproduction. ### MAD Self-Play Score: 101.3. Result ID: 6a27c7aefbc373592c2458b2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3×8B"}. This record does not assert independent reproduction. ### Role-Play Self-Play Score: 98.4. Result ID: 6a27c7aefbc373592c2458b3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3×8B"}. This record does not assert independent reproduction. ### Qwen3-32B-Inst Self-Play Score: 95.7. Result ID: 6a27c7aefbc373592c2458ae. [Reporting paper]() Recorded evaluation configuration: {"Size":"1×32B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.