minimax-m3.1-flash: 60.7 | Keygen Bench
Rank 36 of 83. Best of 3 attempts writing a keygen tune in FastTracker II. Listen to it in the tracker.
Keygen Bench asks AI models to compose a keygen-style chiptune in FastTracker II, with no network and a Bash tool, three independent attempts each at its highest declared reasoning tier. An AI listener (Meta Audiobox Aesthetics) rates how enjoyable a trusted FT2 render of each tune is, and models are ranked by their best of three attempts.
Ranking, best of 3 attempts| Rank | Model | Maker | Best score | Attempt scores |
|---|
| 1 | gpt-6-sol | OpenAI | 73.7 | 73.7 / 69.6 / 65.6 |
| 2 | gpt-6.1-sol | OpenAI | 71.9 | 71.9 / 71.3 / 65.3 |
| 3 | mimo-v2.6-pro | Xiaomi | 71.7 | 71.7 / 43.8 / 65.8 |
| 4 | deepseek-v4-1-flash | DeepSeek | 70.8 | 62.0 / 70.8 / 61.9 |
| 5 | gpt-6-astra | OpenAI | 70.1 | 70.1 / 67.2 / 67.2 |
| 6 | claude-opus-5 | Anthropic | 69.8 | 69.8 / 65.7 / 68.7 |
| 7 | gemini-4-argon | Google | 68.5 | 66.5 / 68.5 / 65.6 |
| 8 | glm-5.2 | Zhipu | 68.3 | 68.3 / 62.5 / 66.3 |
| 9 | gpt-5.6-sol | OpenAI | 67.2 | 67.2 / 57.0 / 63.0 |
| 10 | claude-fable-5-1 | Anthropic | 67.0 | 64.9 / 67.0 / 64.9 |
| 11 | claude-sonnet-5-5 | Anthropic | 66.5 | 66.5 / 66.1 / 65.7 |
| 12 | claude-sonnet-5 | Anthropic | 66.2 | 66.2 / 61.1 / 45.6 |
| 13 | hy4-preview | Tencent | 65.7 | 42.0 / 61.5 / 65.7 |
| 14 | claude-haiku-5-5 | Anthropic | 65.5 | 57.8 / 65.5 / 58.5 |
| 15 | glm-5-3 | Zhipu | 64.6 | 64.6 / 60.9 / 63.1 |
| 16 | gpt-5-4 | OpenAI | 64.4 | 64.4 / 62.6 / 60.8 |
| 17 | hy3 | Tencent | 64.3 | 31.4 / 38.5 / 64.3 |
| 18 | grok-4-6 | xAI | 64.0 | 64.0 / 51.7 / 54.8 |
| 19 | deepseek-v4-pro | DeepSeek | 63.7 | 63.7 / 54.0 / 35.3 |
| 20 | gpt-5.6-terra | OpenAI | 63.4 | 57.6 / 63.4 / 61.9 |
| 21 | claude-opus-5-5 | Anthropic | 63.2 | 63.2 / 62.5 / 61.4 |
| 22 | mimo-v2.6-flash | Xiaomi | 63.2 | 30.7 / 59.5 / 63.2 |
| 23 | claude-opus-4-7 | Anthropic | 62.5 | 34.5 / 52.1 / 62.5 |
| 24 | glm-5-3-flash | Zhipu | 62.2 | 34.4 / 62.2 |
| 25 | gpt-5.5 | OpenAI | 62.1 | 39.1 / 62.1 / 30.5 |
| 26 | claude-opus-4-8 | Anthropic | 61.9 | 41.8 / 53.4 / 61.9 |
| 27 | gpt-6-luna | OpenAI | 61.9 | 36.8 / 52.4 / 61.9 |
| 28 | kimi-k3 | Moonshot AI | 61.7 | 59.2 / 58.2 / 61.7 |
| 29 | step-5-preview | StepFun | 61.6 | 27.1 / 54.2 / 61.6 |
| 30 | muse-spark-1.3 | Meta | 61.4 | 57.3 / 35.2 / 61.4 |
| 31 | longcat-2.5-preview | Meituan | 61.3 | 51.9 / 61.3 |
| 32 | grok-4-5 | xAI | 61.1 | 31.9 / 28.9 / 61.1 |
| 33 | qwen3.7-plus | Alibaba | 60.8 | 29.7 / 34.1 / 60.8 |
| 34 | deepseek-v4-flash | DeepSeek | 60.7 | 60.7 / 47.0 |
| 35 | gemini-3.8-flash | Google | 60.7 | 37.6 / 57.6 / 60.7 |
| 36 | minimax-m3.1-flash | MiniMax | 60.7 | 40.8 / 60.7 / 50.1 |
| 37 | claude-fable-5 | Anthropic | 60.4 | 60.4 / 52.8 / 49.8 |
| 38 | gemini-3-flash-preview | Google | 60.2 | 60.2 / 31.9 / 31.9 |
| 39 | grok-4-7 | xAI | 59.8 | 59.8 / 41.1 / 50.8 |
| 40 | qwen3.5-397b-a17b | Alibaba | 59.7 | 55.6 / 28.7 / 59.7 |
| 41 | gemini-3.7-flash | Google | 59.3 | 59.3 / 53.9 / 27.8 |
| 42 | claude-sonnet-4-6 | Anthropic | 58.8 | 58.8 / 26.9 / 33.2 |
| 43 | grok-4.20-0309-reasoning | xAI | 57.9 | 47.5 / 44.1 / 57.9 |
| 44 | kimi-k2-6 | Moonshot AI | 57.7 | 42.8 / 57.7 / 39.6 |
| 45 | swe-2 | Cognition | 56.6 | 56.6 / 53.4 / 56.4 |
| 46 | claude-opus-4-6 | Anthropic | 56.3 | 32.3 / 42.8 / 56.3 |
| 47 | grok-composer-2.5-fast | xAI | 55.5 | 55.5 / 38.6 / 52.3 |
| 48 | minimax-m2.7 | MiniMax | 54.3 | 54.3 / 52.3 / 40.9 |
| 49 | gpt-5.6-luna | OpenAI | 53.7 | 53.7 / 33.5 / 24.5 |
| 50 | minimax-m3 | MiniMax | 53.3 | 27.1 / 53.3 / 37.2 |
| 51 | gemini-3.1-pro-preview | Google | 52.2 | 28.1 / 26.0 / 52.2 |
| 52 | gpt-5-4-mini | OpenAI | 52.1 | 26.3 / 52.1 / 34.3 |
| 53 | kimi-k2.7-code | Moonshot AI | 51.7 | 29.3 / 51.7 / 37.2 |
| 54 | command-a-plus-05-2026 | Cohere | 51.5 | 28.7 / 51.5 |
| 55 | gemini-3.5-flash | Google | 51.5 | 51.5 / 50.7 / 24.5 |
| 56 | gpt-5-3-codex | OpenAI | 51.5 | 51.5 / 42.9 / 41.4 |
| 57 | gemini-3.6-flash | Google | 49.5 | 35.9 / 49.5 / 29.4 |
| 58 | hermes-4-405b | Nous Research | 49.0 | 49.0 / 20.5 |
| 59 | claude-opus-4-5-20251101 | Anthropic | 48.8 | 48.8 / 33.5 / 48.1 |
| 60 | swe-1-7-lightning | Cognition | 48.2 | 46.8 / 31.7 / 48.2 |
| 61 | muse-glimmer-30b | Meta | 47.6 | 47.6 / 32.7 / 45.7 |
| 62 | mistral-large-4 | Mistral AI | 47.2 | 34.8 / 47.2 / 35.3 |
| 63 | inkling | Thinking Machines Lab | 47.1 | 47.1 / 40.9 / 23.9 |
| 64 | north-mini-code | Cohere | 47.1 | 47.1 / 25.9 |
| 65 | step-3.7-flash | StepFun | 45.4 | 33.4 / 45.4 / 29.3 |
| 66 | kimi-k2-7 | Moonshot AI | 45.3 | 45.3 / 43.3 / 26.9 |
| 67 | mimo-v2.5 | Xiaomi | 42.4 | 23.6 / 42.4 / 28.7 |
| 68 | qwen3.8-flash | Alibaba | 42.2 | 22.0 / 42.2 / 32.9 |
| 69 | nemotron-3-ultra | NVIDIA | 40.5 | 29.5 / 40.5 / 27.7 |
| 70 | grok-4.3 | xAI | 38.4 | 30.6 / 38.4 / 34.9 |
| 71 | swe-1-6 | Cognition | 37.9 | 18.4 / 37.9 / 20.2 |
| 72 | gpt-oss-120b | OpenAI | 36.8 | 36.8 / 20.5 |
| 73 | claude-haiku-4-5-20251001 | Anthropic | 36.6 | 36.6 / 29.8 / 20.1 |
| 74 | gpt-oss-20b | OpenAI | 36.2 | 36.2 / 34.2 |
| 75 | grok-build-0.1 | xAI | 33.6 | 30.5 / 33.6 / 19.2 |
| 76 | muse-spark-1.2 | Meta | 33.1 | 30.6 / 25.8 / 33.1 |
| 77 | qwen3.8-27b | Alibaba | 32.9 | 32.9 |
| 78 | mercury-2.5 | Inception | 30.7 | 18.8 / 30.7 |
| 79 | grok-4.20-0309-non-reasoning | xAI | 26.6 | 22.4 / 26.6 |
| 80 | grok-build | xAI | 25.3 | 20.3 / 25.3 / 24.9 |
| 81 | nemotron-3-super-120b-a12b | NVIDIA | 24.7 | 22.3 / 24.7 / 22.0 |
| 82 | command-a-reasoning-08-2025 | Cohere | 22.2 | 22.2 |
| 83 | nemotron-3.5-lightning | NVIDIA | 20.7 | 20.7 |
Rankings | Tracker | Scoring | Support