Calibration
AA-SciCode → CursorBench 3.1
CursorBench 3.1 is estimated from AA-SciCode with a linear curve fitted on 5 models measured on both: y = 2.4655·x + -0.7927, R² = 0.91, cross-validated error 3.4 pp. It is used for 100 estimates.
| Estimated model | AA-SciCode | CursorBench 3.1 | Source |
|---|---|---|---|
| A.X K2 | 41.0% | 21.8% | estimated ± 3.4 pp, low confidence |
| Apodex 1.1 | 45.5% | 32.9% | estimated ± 3.4 pp, low confidence |
| Apodex 1.1 Mini | 45.5% | 32.9% | estimated ± 3.4 pp, low confidence |
| Celeris-1 | 21.6% | 0.0% | estimated ± 3.4 pp, low confidence |
| Claude Fable 5.1 | 63.1% | 76.3% | estimated ± 3.4 pp, low confidence |
| Claude Haiku 5.5 | 55.0% | 56.3% | estimated ± 3.4 pp, medium confidence |
| Claude Opus 5 | 56.4% | 59.8% | estimated ± 3.4 pp, medium confidence |
| Claude Opus 5.5 | 66.9% | 85.7% | estimated ± 3.4 pp, low confidence |
| Claude Sonnet 5 | 54.3% | 54.6% | estimated ± 3.4 pp, medium confidence |
| Claude Sonnet 5.5 | 61.0% | 71.1% | estimated ± 3.4 pp, medium confidence |
| Command A+ | 38.5% | 15.6% | estimated ± 3.4 pp, low confidence |
| DeepSeek V3 | 35.8% | 9.0% | estimated ± 3.4 pp, low confidence |
| DeepSeek V3 0324 | 39.0% | 16.9% | estimated ± 3.4 pp, low confidence |
| DeepSeek V4.1 Flash | 51.9% | 48.7% | estimated ± 3.4 pp, medium confidence |
| DeepSeek V4 Flash 0731 | 50.3% | 44.7% | estimated ± 3.4 pp, low confidence |
| DeepSeek V4 Pro 0813 | 51.0% | 46.5% | estimated ± 3.4 pp, low confidence |
| Gemini 2.5 Pro | 46.3% | 34.9% | estimated ± 3.4 pp, low confidence |
| Gemini 3.1 Pro | 58.7% | 65.5% | estimated ± 3.4 pp, medium confidence |
| Gemini 3.5 Flash-Lite | 41.3% | 22.6% | estimated ± 3.4 pp, low confidence |
| Gemini 3.6 Flash | 53.4% | 52.4% | estimated ± 3.4 pp, medium confidence |
| Gemini 3.7 Flash | 57.2% | 61.8% | estimated ± 3.4 pp, medium confidence |
| Gemini 3.8 Flash | 56.6% | 60.3% | estimated ± 3.4 pp, medium confidence |
| Gemini 4 Argon | 61.8% | 73.1% | estimated ± 3.4 pp, low confidence |
| Gemma 3 27B | 23.3% | 0.0% | estimated ± 3.4 pp, low confidence |
| Gemma 4 26B A4B | 40.0% | 19.3% | estimated ± 3.4 pp, low confidence |
| Gemma 4 31B | 45.5% | 32.9% | estimated ± 3.4 pp, low confidence |
| Gemma 4 E4B | 24.4% | 0.0% | estimated ± 3.4 pp, low confidence |
| GLM-5.1 | 44.8% | 31.2% | estimated ± 3.4 pp, low confidence |
| GLM-5.2 | 51.2% | 47.0% | estimated ± 3.4 pp, low confidence |
| GLM-5.3 | 59.0% | 66.2% | estimated ± 3.4 pp, medium confidence |
| GLM-5.3-Flash | 51.6% | 47.9% | estimated ± 3.4 pp, medium confidence |
| GPT-5.4 mini | 52.1% | 49.2% | estimated ± 3.4 pp, medium confidence |
| GPT-5.4 nano | 47.2% | 37.1% | estimated ± 3.4 pp, low confidence |
| GPT-5.6 Luna | 53.6% | 52.9% | estimated ± 3.4 pp, medium confidence |
| GPT-5.6 Sol | 57.1% | 61.5% | estimated ± 3.4 pp, medium confidence |
| GPT-5.6 Terra | 55.0% | 56.3% | estimated ± 3.4 pp, medium confidence |
| GPT-6.1 Sol | 54.2% | 54.4% | estimated ± 3.4 pp, medium confidence |
| GPT-6 Astra | 56.5% | 60.0% | estimated ± 3.4 pp, medium confidence |
| GPT-6 Luna | 54.6% | 55.3% | estimated ± 3.4 pp, medium confidence |
| GPT-6 Sol | 57.6% | 62.7% | estimated ± 3.4 pp, medium confidence |
| GPT-OSS 120B | 34.0% | 4.6% | estimated ± 3.4 pp, low confidence |
| GPT-OSS 20B | 38.9% | 16.6% | estimated ± 3.4 pp, low confidence |
| Granite 4.2 30B | 37.8% | 13.9% | estimated ± 3.4 pp, low confidence |
| Granite 4.2 3B | 25.3% | 0.0% | estimated ± 3.4 pp, low confidence |
| Granite 4.2 8B | 31.5% | 0.0% | estimated ± 3.4 pp, low confidence |
| Grok 4.3 | 48.3% | 39.8% | estimated ± 3.4 pp, low confidence |
| Grok 4.5 | 55.0% | 56.3% | estimated ± 3.4 pp, medium confidence |
| Grok 4.6 | 56.5% | 60.0% | estimated ± 3.4 pp, medium confidence |
| Grok 4.7 | 57.4% | 62.2% | estimated ± 3.4 pp, medium confidence |
| Hy3 | 48.6% | 40.6% | estimated ± 3.4 pp, low confidence |
| Hy3 Preview | 48.6% | 40.6% | estimated ± 3.4 pp, low confidence |
| Inkling | 47.0% | 36.6% | estimated ± 3.4 pp, low confidence |
| Inkling-Small | 49.7% | 43.3% | estimated ± 3.4 pp, low confidence |
| K-EXAONE 2.0 | 42.0% | 24.3% | estimated ± 3.4 pp, low confidence |
| Kimi K2.7 Code | 47.8% | 38.6% | estimated ± 3.4 pp, low confidence |
| Kimi K3 | 59.5% | 67.4% | estimated ± 3.4 pp, medium confidence |
| LFM2.5-2.6B | 14.4% | 0.0% | estimated ± 3.4 pp, low confidence |
| Ling 3.0 Flash | 42.0% | 24.3% | estimated ± 3.4 pp, low confidence |
| Ling 3.0 Flash FP8 | 42.0% | 24.3% | estimated ± 3.4 pp, low confidence |
| Ling 3.0 Flash VL | 44.2% | 29.7% | estimated ± 3.4 pp, low confidence |
| Ling 3.0 Tiny | 24.2% | 0.0% | estimated ± 3.4 pp, low confidence |
| Ling 3.1 Flash | 54.1% | 54.1% | estimated ± 3.4 pp, medium confidence |
| Llama 4 Maverick | 31.7% | 0.0% | estimated ± 3.4 pp, low confidence |
| Llama 4 Scout | 21.3% | 0.0% | estimated ± 3.4 pp, low confidence |
| Mercury 2.5 | 38.5% | 15.6% | estimated ± 3.4 pp, low confidence |
| MiMo-V2.5-Pro | 50.6% | 45.5% | estimated ± 3.4 pp, low confidence |
| MiMo-V2.6-Flash | 51.3% | 47.2% | estimated ± 3.4 pp, low confidence |
| MiMo-V2.6-Pro | 60.9% | 70.9% | estimated ± 3.4 pp, medium confidence |
| MiniCPM5-2B | 26.3% | 0.0% | estimated ± 3.4 pp, low confidence |
| MiniMax M2.7 | 50.1% | 44.3% | estimated ± 3.4 pp, low confidence |
| MiniMax M3 | 47.1% | 36.9% | estimated ± 3.4 pp, low confidence |
| Mistral Large 3 | 36.6% | 11.0% | estimated ± 3.4 pp, low confidence |
| Mistral Large 4 | 54.2% | 54.4% | estimated ± 3.4 pp, medium confidence |
| Mistral Medium 3.5 128B | 40.2% | 19.8% | estimated ± 3.4 pp, low confidence |
| Mistral Small 4 | 38.8% | 16.4% | estimated ± 3.4 pp, low confidence |
| Mistral Small 4 (Reasoning) | 38.8% | 16.4% | estimated ± 3.4 pp, low confidence |
| Muse Glimmer 30B | 44.9% | 31.4% | estimated ± 3.4 pp, low confidence |
| Muse Spark 1.1 | 58.8% | 65.7% | estimated ± 3.4 pp, medium confidence |
| Muse Spark 1.2 | 57.4% | 62.2% | estimated ± 3.4 pp, medium confidence |
| Muse Spark 1.3 | 58.8% | 65.7% | estimated ± 3.4 pp, medium confidence |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 32.1% | 0.0% | estimated ± 3.4 pp, low confidence |
| Nemotron 3 Nano 30B | 30.6% | 0.0% | estimated ± 3.4 pp, low confidence |
| Nemotron 3 Super 100B | 36.2% | 10.0% | estimated ± 3.4 pp, low confidence |
| Nemotron 3 Ultra | 40.3% | 20.1% | estimated ± 3.4 pp, low confidence |
| North Mini Code | 38.8% | 16.4% | estimated ± 3.4 pp, low confidence |
| Quasar 438B | 48.1% | 39.3% | estimated ± 3.4 pp, low confidence |
| Qwen3.5-122B-A10B | 39.7% | 18.6% | estimated ± 3.4 pp, low confidence |
| Qwen3.6-27B | 42.8% | 26.3% | estimated ± 3.4 pp, low confidence |
| Qwen3.6-35B-A3B | 36.6% | 11.0% | estimated ± 3.4 pp, low confidence |
| Qwen3.7 Max | 49.5% | 42.8% | estimated ± 3.4 pp, low confidence |
| Qwen3.7 Plus | 46.1% | 34.4% | estimated ± 3.4 pp, low confidence |
| Qwen3.8-27B | 46.6% | 35.6% | estimated ± 3.4 pp, low confidence |
| Qwen3.8-Flash-Next | 50.6% | 45.5% | estimated ± 3.4 pp, low confidence |
| Qwen3.8 Max Preview | 52.1% | 49.2% | estimated ± 3.4 pp, medium confidence |
| Solar Pro 3 | 25.5% | 0.0% | estimated ± 3.4 pp, low confidence |
| Solar Pro 4 | 44.6% | 30.7% | estimated ± 3.4 pp, low confidence |
| Step 3.7 Flash | 43.9% | 29.0% | estimated ± 3.4 pp, low confidence |
| Step 5 Preview | 58.9% | 65.9% | estimated ± 3.4 pp, medium confidence |
| Trinity-Large-Preview | 40.6% | 20.8% | estimated ± 3.4 pp, low confidence |
| Trinity-Large-Thinking | 40.6% | 20.8% | estimated ± 3.4 pp, low confidence |