Benchmark scores for recent small open-weight AI models (4B–190B; official weights, current versions only).
Generated 2026-09-05T19:53:15Z
Click any column header to sort (Params / Benchmarks / AA-Index sort numerically, Released by date; click again to flip direction; missing values sink to the bottom). A V badge marks models that support vision / multimodal (image) input.
| Model | Family | Params | Architecture | Released | Benchmarks | AA-Index |
|---|---|---|---|---|---|---|
| Qwen3.8-27BV | Qwen3.8 | 27.8B | dense | 2026-08-05 | 26 | 41 |
| Qwen3.8-Flash-NextV | Qwen3.8 | 180B | moe | 2026-08-24 | 23 | 46 |
| Qwen-AgentWorld-35B-A3BV | Qwen-AgentWorld | 34.7B | moe | 2026-06-22 | 1 | — |
| gemma-4-12B-itV | gemma-4 | 12B | dense | 2026-05-23 | 17 | 16 |
| diffusiongemma-26B-A4B-itV | diffusiongemma | 25.8B | moe | 2026-06-09 | 15 | 8 |
| DeepSeek-V4-Flash-DSpark | DeepSeek-V4 | 165.3B | moe | 2026-06-27 | 22 | 41 |
| Laguna-S-2.1 | Laguna | 117.6B | moe | 2026-07-13 | 6 | — |
| Laguna-XS-2.1 | Laguna | 33.4B | moe | 2026-06-20 | 4 | — |
| LFM2.5-8B-A1B | LFM2.5 | 8.5B | moe | 2026-05-28 | 11 | 3 |
| Nanbeige4.2-3B | Nanbeige4.2 | 4.2B | dense | 2026-07-21 | 21 | — |
| KAT-Coder-V2.5-DevV | KAT-Coder | 34.7B | dense | 2026-07-23 | 7 | — |
| North-Mini-Code-1.0 | North | 30.5B | moe | 2026-06-05 | 9 | 13.4 |
| Mellum2-12B-A2.5B-Thinking | Mellum2 | 12.1B | moe | 2026-05-26 | 12 | — |
| granite-4.2-30b | granite-4.2 | 29.3B | dense | 2026-08-07 | 20 | 17 |
| granite-4.2-8b | granite-4.2 | 8.8B | dense | 2026-08-07 | 20 | 14 |
| Hy-MT2-30B-A3B | Hy-MT2 | 30.1B | moe | 2026-05-11 | 9 | — |
| LongCat-Flash-Lite-Sparse | LongCat | 69.1B | dense | 2026-07-31 | 30 | 11 |
| Fara1.5-27BV | Fara | 27.4B | dense | 2026-07-17 | 3 | — |
| Fara1.5-9BV | Fara | 9.4B | dense | 2026-05-12 | 3 | — |
| Nemotron-3-Nano-Omni-30B-A3B-ReasoningV | Nemotron-3 | 33B | moe | 2026-04-20 | 12 | 9 |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | Nemotron-3.5 | 31.6B | moe | 2026-08-01 | 15 | 16 |
| Ornith-1.5-9BV | Ornith | 9.7B | dense | 2026-08-18 | 13 | — |
| Ornith-1.5-35B-A3BV | Ornith | 36B | moe | 2026-08-18 | 15 | — |
| Ling-3.0-tiny | Ling | 7.9B | moe | 2026-08-10 | 18 | 25 |
| Ling-3.0-flash | Ling | 127.5B | moe | 2026-08-02 | 26 | 27 |
| Spark-X2.5-4B | Spark X | 4.1B | dense | 2026-08-24 | 21 | — |
Cell shows the score for that model×benchmark; grey = absent / no numeric score. Cell colour is scaled per benchmark (light → deep blue = lower → higher); the gold cell is the best model on that benchmark. Columns are ordered by how many models report each benchmark. Click a benchmark column to rank models by it; click the Model column to sort by name.
| Model | GPQA Diamond | SWE-bench Pro | Terminal-Bench | AA-Intelligence-… | HLE | SWE-bench Multil… | IFBench | SWE-bench Verifi… | LiveCodeBench | AIME | SciCode | BFCL | BrowseComp | MCP-Atlas | MMLU-Pro | AA-LCR | HMMT Feb 2026 | IMO AnswerBench | DeepSWE | IFEval | MathVision | NL2Repo-Bench | Toolathlon | τ²-bench | ClawEval | Codeforces ELO | GDPval | GDPval-AA | MRCR | Multi-IF | OmniDocBench 1.5 | WideSearch | τ³-bench |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Qwen3.8-27BV | 89.2 | 61.7 | 73 | 41 | 30.8 | · | 79.5 | · | 90.3 | · | · | · | · | · | · | · | · | · | 42.2 | · | 90 | 42.3 | · | · | · | · | · | · | · | · | 91.1 | · | · |
| Qwen3.8-Flash-NextV | 91.7 | 62.5 | · | 46 | 35.9 | 81 | 81.3 | · | 91.9 | · | · | · | · | · | · | · | · | · | 58.7 | · | 90.6 | 48.1 | 73.5 | · | · | · | · | · | · | · | · | · | · |
| Qwen-AgentWorld-35B-A3BV | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| gemma-4-12B-itV | 78.8 | · | · | 16 | 5.2 | · | · | · | 72 | 77.5 | · | · | · | · | 77.2 | · | · | · | · | · | 79.7 | · | · | 69 | · | 1659 | · | · | · | · | 0.2 | · | · |
| diffusiongemma-26B-A4B-itV | 73.2 | · | · | 8 | 11 | · | · | · | 69.1 | 69.1 | · | · | · | · | 77.6 | · | · | · | · | · | 70.5 | · | · | 56.2 | · | 1429 | · | · | · | · | 0.3 | · | · |
| DeepSeek-V4-Flash-DSpark | 88.1 | 52.6 | 56.9 | 41 | 34.8 | 73.3 | · | 79 | 91.6 | · | · | · | 73.2 | 69 | 86.2 | · | 94.8 | 88.4 | · | · | · | · | 47.8 | · | · | 3052 | · | 1395 | 78.7 | · | · | · | · |
| Laguna-S-2.1 | · | 59.4 | 70.2 | · | · | 78.5 | · | · | · | · | · | · | · | · | · | · | · | · | 40.4 | · | · | · | 49.7 | · | · | · | · | · | · | · | · | · | · |
| Laguna-XS-2.1 | · | 47.6 | 37.5 | · | · | 63.1 | · | 70.9 | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| LFM2.5-8B-A1B | · | · | · | 3 | · | · | 56.5 | · | · | 42.5 | · | 64.8 | · | · | · | · | · | · | · | 91.8 | · | · | · | 88.1 | · | · | · | · | · | 79.9 | · | · | · |
| Nanbeige4.2-3B | 87.4 | 46.9 | 44.1 | · | 17.8 | · | 54.6 | 63.6 | 72.5 | · | 35.6 | · | · | 57.8 | · | 58.7 | 82.8 | 67.3 | · | · | · | · | · | · | 52.2 | · | 74.3 | · | · | · | · | · | · |
| KAT-Coder-V2.5-DevV | · | 46.0 | 41.0 | · | · | 63 | · | 69.4 | · | · | 44.2 | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| North-Mini-Code-1.0 | 75.7 | 40.2 | 36 | 13.4 | 11.1 | · | 57.5 | 67.6 | 70.3 | · | 38.2 | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| Mellum2-12B-A2.5B-Thinking | 57.6 | · | · | · | · | · | · | · | 69.9 | 58.4 | · | 45.6 | · | · | · | · | · | · | · | 76.5 | · | · | · | · | · | · | · | · | · | · | · | · | · |
| granite-4.2-30b | 66.4 | 33.3 | 29.2 | 17 | · | 41.9 | 77.2 | 57 | 75.8 | 89.2 | 38.8 | 61.4 | · | · | 77.6 | · | · | · | · | · | · | · | · | · | · | · | 1225 | · | · | · | · | · | 62 |
| granite-4.2-8b | 64.1 | 19.1 | 20.6 | 14 | · | 30.8 | 79.3 | 47.7 | 73.2 | 86.7 | 36.1 | 52.4 | · | · | 74.0 | · | · | · | · | · | · | · | · | · | · | · | 1189 | · | · | · | · | · | 58.1 |
| Hy-MT2-30B-A3B | · | · | · | · | · | · | 50.7 | · | · | · | · | · | · | · | · | · | · | · | · | 89.8 | · | · | · | · | · | · | · | · | · | · | · | · | · |
| LongCat-Flash-Lite-Sparse | 69.5 | 40.6 | 33.7 | 11 | · | 59.3 | · | 68.2 | · | 65.7 | · | · | 48.6 | 45.6 | 79.2 | 48 | 40.5 | 49.4 | · | · | · | · | · | · | · | · | · | · | 44.7 | · | · | · | · |
| Fara1.5-27BV | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| Fara1.5-9BV | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| Nemotron-3-Nano-Omni-30B-A3B-ReasoningV | · | · | · | 9 | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 75.4 | · | 24.6 | 16 | 11.7 | 39.3 | 71.9 | 51.6 | · | · | 32.6 | · | 37.0 | · | 81.9 | 52 | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · |
| Ornith-1.5-9BV | 86.4 | 47.5 | 46.2 | · | 20.2 | 54.4 | · | 70.6 | · | · | · | · | 56.4 | 54.2 | · | · | · | · | · | · | · | 32.4 | · | · | 66.5 | · | · | · | · | · | · | 59.5 | · |
| Ornith-1.5-35B-A3BV | 89.2 | 59.6 | 67.8 | · | 25.6 | 71.4 | · | 79 | · | · | · | · | 67.6 | 70.2 | · | · | · | · | 22 | · | · | 46.2 | 48.7 | · | 72.5 | · | · | · | · | · | · | 67.8 | · |
| Ling-3.0-tiny | 73.4 | · | 27.7 | 25 | 9.3 | · | 63.6 | · | · | · | 24.2 | 62.7 | · | · | · | 58.7 | 70.3 | 71.0 | · | · | · | · | · | · | · | · | · | 772 | · | 83.2 | · | · | · |
| Ling-3.0-flash | · | 56.6 | 57 | 27 | 22.7 | 72.4 | 74.5 | · | 82.8 | 93.2 | · | 73 | 72.2 | 65.5 | · | 65.1 | 87 | 83.7 | · | · | · | · | · | · | · | · | · | 1107 | 90.8 | 87.7 | · | 73.6 | · |
| Spark-X2.5-4B | 67.4 | 44.4 | · | · | 12.3 | 53.3 | 75 | 41.6 | · | 90.7 | 34.7 | 65.1 | 40.9 | 54.6 | · | 56.3 | 81.2 | 74.2 | · | 93 | · | · | · | 75.1 | · | · | · | · | · | · | · | · | 30.4 |
Bar charts compare models that report the same benchmark (≥2 models within the parameter range above). Hover a bar for the exact value.
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 91.7 | % | Hugging Face model card | link | |
| Qwen3.8-27B | 89.2 | % | Hugging Face model card | link | |
| Ornith-1.5-35B-A3B | 89.2 | % | Hugging Face model card | link | |
| DeepSeek-V4-Flash-DSpark | 88.1 | % | Pass@1 | Hugging Face model card | link |
| Nanbeige4.2-3B | 87.4 | % | Hugging Face model card | link | |
| Ornith-1.5-9B | 86.4 | % | Hugging Face model card | link | |
| gemma-4-12B-it | 78.8 | % | Hugging Face model card | link | |
| North-Mini-Code-1.0 | 75.7 | % | Overall | Artificial Analysis | link |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 75.4 | % | no tools | Hugging Face model card | link |
| Ling-3.0-tiny | 73.4 | % | Hugging Face model card | link | |
| diffusiongemma-26B-A4B-it | 73.2 | % | Hugging Face model card | link | |
| LongCat-Flash-Lite-Sparse | 69.5 | % | avg@16, w/o HI | Hugging Face model card | link |
| Spark-X2.5-4B | 67.4 | % | Hugging Face model card | link | |
| granite-4.2-30b | 66.4 | % | Hugging Face model card | link | |
| granite-4.2-8b | 64.1 | % | Hugging Face model card | link | |
| Mellum2-12B-A2.5B-Thinking | 57.6 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 62.5 | % | Hugging Face model card | link | |
| Qwen3.8-27B | 61.7 | % | Hugging Face model card | link | |
| Ornith-1.5-35B-A3B | 59.6 | % | Hugging Face model card | link | |
| Laguna-S-2.1 | 59.4 | % | Public Dataset | Hugging Face model card | link |
| Ling-3.0-flash | 56.6 | % | Hugging Face model card | link | |
| DeepSeek-V4-Flash-DSpark | 52.6 | % | Resolved | Hugging Face model card | link |
| Laguna-XS-2.1 | 47.6 | % | Public Dataset, pass@1 (avg over 2 attempts) | Hugging Face model card | link |
| Ornith-1.5-9B | 47.5 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 46.9 | % | Hugging Face model card | link | |
| KAT-Coder-V2.5-Dev | 46.0 | % | pass@1 | Hugging Face model card | link |
| Spark-X2.5-4B | 44.4 | % | Hugging Face model card | link | |
| LongCat-Flash-Lite-Sparse | 40.6 | % | w/o HI | Hugging Face model card | link |
| North-Mini-Code-1.0 | 40.2 | % | Overall | Hugging Face model card | link |
| granite-4.2-30b | 33.3 | % | Hugging Face model card | link | |
| granite-4.2-8b | 19.1 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-27B | 73 | % | 2.1 (Terminus) | Hugging Face model card | link |
| Laguna-S-2.1 | 70.2 | % | 2.1 | Hugging Face model card | link |
| Ornith-1.5-35B-A3B | 67.8 | % | 2.1 (Terminus-2) | Hugging Face model card | link |
| Ling-3.0-flash | 57 | % | 2.1 | Hugging Face model card | link |
| DeepSeek-V4-Flash-DSpark | 56.9 | % | 2.0 | Hugging Face model card | link |
| Ornith-1.5-9B | 46.2 | % | 2.1 (Terminus-2) | Hugging Face model card | link |
| Nanbeige4.2-3B | 44.1 | % | 2.0 | Hugging Face model card | link |
| KAT-Coder-V2.5-Dev | 41.0 | % | v2.1, avg of two harnesses | Hugging Face model card | link |
| Laguna-XS-2.1 | 37.5 | % | v2.0, pass@1 (avg over 5 attempts) | Hugging Face model card | link |
| North-Mini-Code-1.0 | 36 | % | v2 | Hugging Face model card | link |
| LongCat-Flash-Lite-Sparse | 33.7 | % | 2.0, w/o HI | Hugging Face model card | link |
| granite-4.2-30b | 29.2 | % | 2.1 | Hugging Face model card | link |
| Ling-3.0-tiny | 27.7 | % | 2.1 | Hugging Face model card | link |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 24.6 | % | v2.1 | Hugging Face model card | link |
| granite-4.2-8b | 20.6 | % | 2.1 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 46 | idx | Artificial Analysis | link | |
| Qwen3.8-27B | 41 | index | reasoning level: xhigh (highest) | Artificial Analysis | link |
| DeepSeek-V4-Flash-DSpark | 41 | idx | Artificial Analysis | link | |
| Ling-3.0-flash | 27 | idx | Artificial Analysis | link | |
| Ling-3.0-tiny | 25 | idx | v4.1.1 | Hugging Face model card | link |
| granite-4.2-30b | 17 | index | Artificial Analysis | link | |
| gemma-4-12B-it | 16 | index | Artificial Analysis | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 16 | index | Artificial Analysis | link | |
| granite-4.2-8b | 14 | index | Artificial Analysis | link | |
| North-Mini-Code-1.0 | 13.4 | idx | v4.2 | Artificial Analysis | link |
| LongCat-Flash-Lite-Sparse | 11 | index | Artificial Analysis | link | |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning | 9 | index | Artificial Analysis | link | |
| diffusiongemma-26B-A4B-it | 8 | index | Artificial Analysis | link | |
| LFM2.5-8B-A1B | 3 | idx | Artificial Analysis | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 35.9 | % | Hugging Face model card | link | |
| DeepSeek-V4-Flash-DSpark | 34.8 | % | Pass@1 | Hugging Face model card | link |
| Qwen3.8-27B | 30.8 | % | Hugging Face model card | link | |
| Ornith-1.5-35B-A3B | 25.6 | % | no tools | Hugging Face model card | link |
| Ling-3.0-flash | 22.7 | % | Hugging Face model card | link | |
| Ornith-1.5-9B | 20.2 | % | no tools | Hugging Face model card | link |
| Nanbeige4.2-3B | 17.8 | % | w/o search | Hugging Face model card | link |
| Spark-X2.5-4B | 12.3 | % | Hugging Face model card | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 11.7 | % | text-only, no tools | Hugging Face model card | link |
| North-Mini-Code-1.0 | 11.1 | % | Overall | Artificial Analysis | link |
| diffusiongemma-26B-A4B-it | 11 | % | no tools | Hugging Face model card | link |
| Ling-3.0-tiny | 9.3 | % | Hugging Face model card | link | |
| gemma-4-12B-it | 5.2 | % | no tools | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 81 | % | Hugging Face model card | link | |
| Laguna-S-2.1 | 78.5 | % | Hugging Face model card | link | |
| DeepSeek-V4-Flash-DSpark | 73.3 | % | Resolved | Hugging Face model card | link |
| Ling-3.0-flash | 72.4 | % | Hugging Face model card | link | |
| Ornith-1.5-35B-A3B | 71.4 | % | Hugging Face model card | link | |
| Laguna-XS-2.1 | 63.1 | % | pass@1 (avg over 4 attempts) | Hugging Face model card | link |
| KAT-Coder-V2.5-Dev | 63 | % | pass@1 | Hugging Face model card | link |
| LongCat-Flash-Lite-Sparse | 59.3 | % | w/o HI | Hugging Face model card | link |
| Ornith-1.5-9B | 54.4 | % | Hugging Face model card | link | |
| Spark-X2.5-4B | 53.3 | % | Hugging Face model card | link | |
| granite-4.2-30b | 41.9 | % | Hugging Face model card | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 39.3 | % | Overall | Hugging Face model card | link |
| granite-4.2-8b | 30.8 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 81.3 | % | Hugging Face model card | link | |
| Qwen3.8-27B | 79.5 | % | Hugging Face model card | link | |
| granite-4.2-8b | 79.3 | % | prompt | Hugging Face model card | link |
| granite-4.2-30b | 77.2 | % | prompt | Hugging Face model card | link |
| Spark-X2.5-4B | 75 | % | Hugging Face model card | link | |
| Ling-3.0-flash | 74.5 | % | Hugging Face model card | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 71.9 | % | loose | Hugging Face model card | link |
| Ling-3.0-tiny | 63.6 | % | Hugging Face model card | link | |
| North-Mini-Code-1.0 | 57.5 | % | Overall | Artificial Analysis | link |
| LFM2.5-8B-A1B | 56.5 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 54.6 | % | Hugging Face model card | link | |
| Hy-MT2-30B-A3B | 50.7 | % | Overall | Official tech report / arXiv paper | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-DSpark | 79 | % | Resolved | Hugging Face model card | link |
| Ornith-1.5-35B-A3B | 79 | % | Hugging Face model card | link | |
| Laguna-XS-2.1 | 70.9 | % | pass@1 (avg over 4 attempts) | Hugging Face model card | link |
| Ornith-1.5-9B | 70.6 | % | Hugging Face model card | link | |
| KAT-Coder-V2.5-Dev | 69.4 | % | pass@1 | Hugging Face model card | link |
| LongCat-Flash-Lite-Sparse | 68.2 | % | w/o HI | Hugging Face model card | link |
| North-Mini-Code-1.0 | 67.6 | % | Overall | Hugging Face model card | link |
| Nanbeige4.2-3B | 63.6 | % | Hugging Face model card | link | |
| granite-4.2-30b | 57 | % | Hugging Face model card | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 51.6 | % | Overall | Hugging Face model card | link |
| granite-4.2-8b | 47.7 | % | Hugging Face model card | link | |
| Spark-X2.5-4B | 41.6 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 91.9 | % | v6 | Hugging Face model card | link |
| DeepSeek-V4-Flash-DSpark | 91.6 | % | Pass@1 | Hugging Face model card | link |
| Qwen3.8-27B | 90.3 | % | v6 | Hugging Face model card | link |
| Ling-3.0-flash | 82.8 | % | 2408-2505 | Hugging Face model card | link |
| granite-4.2-30b | 75.8 | % | v6 | Hugging Face model card | link |
| granite-4.2-8b | 73.2 | % | v6 | Hugging Face model card | link |
| Nanbeige4.2-3B | 72.5 | % | v6 | Hugging Face model card | link |
| gemma-4-12B-it | 72 | % | v6 | Hugging Face model card | link |
| North-Mini-Code-1.0 | 70.3 | % | v6 | Hugging Face model card | link |
| Mellum2-12B-A2.5B-Thinking | 69.9 | % | v6 | Hugging Face model card | link |
| diffusiongemma-26B-A4B-it | 69.1 | % | v6 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ling-3.0-flash | 93.2 | % | 2026 | Hugging Face model card | link |
| Spark-X2.5-4B | 90.7 | % | 2026 | Hugging Face model card | link |
| granite-4.2-30b | 89.2 | % | 2025 | Hugging Face model card | link |
| granite-4.2-8b | 86.7 | % | 2025 | Hugging Face model card | link |
| gemma-4-12B-it | 77.5 | % | 2026 no tools | Hugging Face model card | link |
| diffusiongemma-26B-A4B-it | 69.1 | % | 2026 no tools | Hugging Face model card | link |
| LongCat-Flash-Lite-Sparse | 65.7 | % | 2026, avg@32, w/o HI | Hugging Face model card | link |
| Mellum2-12B-A2.5B-Thinking | 58.4 | % | Hugging Face model card | link | |
| LFM2.5-8B-A1B | 42.5 | % | 2025 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| KAT-Coder-V2.5-Dev | 44.2 | % | pass@1 | Hugging Face model card | link |
| granite-4.2-30b | 38.8 | % | Hugging Face model card | link | |
| North-Mini-Code-1.0 | 38.2 | % | Overall | Hugging Face model card | link |
| granite-4.2-8b | 36.1 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 35.6 | % | Hugging Face model card | link | |
| Spark-X2.5-4B | 34.7 | % | Hugging Face model card | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 32.6 | % | Overall | Hugging Face model card | link |
| Ling-3.0-tiny | 24.2 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ling-3.0-flash | 73 | % | v4 | Hugging Face model card | link |
| Spark-X2.5-4B | 65.1 | % | v4 | Hugging Face model card | link |
| LFM2.5-8B-A1B | 64.8 | % | v3 | Hugging Face model card | link |
| Ling-3.0-tiny | 62.7 | % | v4 | Hugging Face model card | link |
| granite-4.2-30b | 61.4 | % | v4 | Hugging Face model card | link |
| granite-4.2-8b | 52.4 | % | v4 | Hugging Face model card | link |
| Mellum2-12B-A2.5B-Thinking | 45.6 | % | v4 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-DSpark | 73.2 | % | Pass@1 | Hugging Face model card | link |
| Ling-3.0-flash | 72.2 | % | single-agent (w/ ctx) | Hugging Face model card | link |
| Ornith-1.5-35B-A3B | 67.6 | % | Hugging Face model card | link | |
| Ornith-1.5-9B | 56.4 | % | Hugging Face model card | link | |
| LongCat-Flash-Lite-Sparse | 48.6 | % | pass@1, w/o HI | Hugging Face model card | link |
| Spark-X2.5-4B | 40.9 | % | Hugging Face model card | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 37.0 | % | Overall | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ornith-1.5-35B-A3B | 70.2 | % | Hugging Face model card | link | |
| DeepSeek-V4-Flash-DSpark | 69 | % | Pass@1 | Hugging Face model card | link |
| Ling-3.0-flash | 65.5 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 57.8 | % | Hugging Face model card | link | |
| Spark-X2.5-4B | 54.6 | % | Hugging Face model card | link | |
| Ornith-1.5-9B | 54.2 | % | Hugging Face model card | link | |
| LongCat-Flash-Lite-Sparse | 45.6 | % | w/o HI | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-DSpark | 86.2 | % | EM | Hugging Face model card | link |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 81.9 | % | Overall | Hugging Face model card | link |
| LongCat-Flash-Lite-Sparse | 79.2 | % | w/o HI | Hugging Face model card | link |
| diffusiongemma-26B-A4B-it | 77.6 | % | Hugging Face model card | link | |
| granite-4.2-30b | 77.6 | % | Hugging Face model card | link | |
| gemma-4-12B-it | 77.2 | % | Hugging Face model card | link | |
| granite-4.2-8b | 74.0 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ling-3.0-flash | 65.1 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 58.7 | % | Hugging Face model card | link | |
| Ling-3.0-tiny | 58.7 | % | Hugging Face model card | link | |
| Spark-X2.5-4B | 56.3 | % | Hugging Face model card | link | |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 | 52 | idx | Overall | Hugging Face model card | link |
| LongCat-Flash-Lite-Sparse | 48 | % | w/o HI | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-DSpark | 94.8 | % | Pass@1 | Hugging Face model card | link |
| Ling-3.0-flash | 87 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 82.8 | % | Hugging Face model card | link | |
| Spark-X2.5-4B | 81.2 | % | Hugging Face model card | link | |
| Ling-3.0-tiny | 70.3 | % | Hugging Face model card | link | |
| LongCat-Flash-Lite-Sparse | 40.5 | % | avg@32, w/o HI | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-DSpark | 88.4 | % | Pass@1 | Hugging Face model card | link |
| Ling-3.0-flash | 83.7 | % | Hugging Face model card | link | |
| Spark-X2.5-4B | 74.2 | % | Hugging Face model card | link | |
| Ling-3.0-tiny | 71.0 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 67.3 | % | Hugging Face model card | link | |
| LongCat-Flash-Lite-Sparse | 49.4 | % | avg@4, w/o HI | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 58.7 | % | 1.1 | Hugging Face model card | link |
| Qwen3.8-27B | 42.2 | % | 1.1 | Hugging Face model card | link |
| Laguna-S-2.1 | 40.4 | % | Hugging Face model card | link | |
| Ornith-1.5-35B-A3B | 22 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Spark-X2.5-4B | 93 | % | Hugging Face model card | link | |
| LFM2.5-8B-A1B | 91.8 | % | Hugging Face model card | link | |
| Hy-MT2-30B-A3B | 89.8 | % | Overall | Official tech report / arXiv paper | link |
| Mellum2-12B-A2.5B-Thinking | 76.5 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 90.6 | % | without CI | Hugging Face model card | link |
| Qwen3.8-27B | 90 | % | without CI | Hugging Face model card | link |
| gemma-4-12B-it | 79.7 | % | Hugging Face model card | link | |
| diffusiongemma-26B-A4B-it | 70.5 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 48.1 | % | Hugging Face model card | link | |
| Ornith-1.5-35B-A3B | 46.2 | % | Hugging Face model card | link | |
| Qwen3.8-27B | 42.3 | % | Hugging Face model card | link | |
| Ornith-1.5-9B | 32.4 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-Flash-Next | 73.5 | % | Verified, Pass@1 | Hugging Face model card | link |
| Laguna-S-2.1 | 49.7 | % | Verified | Hugging Face model card | link |
| Ornith-1.5-35B-A3B | 48.7 | % | Verified | Hugging Face model card | link |
| DeepSeek-V4-Flash-DSpark | 47.8 | % | Pass@1 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| LFM2.5-8B-A1B | 88.1 | % | Telecom | Hugging Face model card | link |
| Spark-X2.5-4B | 75.1 | % | Hugging Face model card | link | |
| gemma-4-12B-it | 69 | % | average over 3 | Hugging Face model card | link |
| diffusiongemma-26B-A4B-it | 56.2 | % | average over 3 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ornith-1.5-35B-A3B | 72.5 | % | Hugging Face model card | link | |
| Ornith-1.5-9B | 66.5 | % | Hugging Face model card | link | |
| Nanbeige4.2-3B | 52.2 | % | pass^3 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-DSpark | 3052 | rating | Hugging Face model card | link | |
| gemma-4-12B-it | 1659 | rating | Hugging Face model card | link | |
| diffusiongemma-26B-A4B-it | 1429 | rating | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| granite-4.2-30b | 1225 | score | Hugging Face model card | link | |
| granite-4.2-8b | 1189 | points | Hugging Face model card | link | |
| Nanbeige4.2-3B | 74.3 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-DSpark | 1395 | Elo | Hugging Face model card | link | |
| Ling-3.0-flash | 1107 | points | v2 | Hugging Face model card | link |
| Ling-3.0-tiny | 772 | points | v2 | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ling-3.0-flash | 90.8 | % | 128K | Hugging Face model card | link |
| DeepSeek-V4-Flash-DSpark | 78.7 | % | 1M | Hugging Face model card | link |
| LongCat-Flash-Lite-Sparse | 44.7 | % | 8-needle, w/o HI | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ling-3.0-flash | 87.7 | % | Hugging Face model card | link | |
| Ling-3.0-tiny | 83.2 | % | Hugging Face model card | link | |
| LFM2.5-8B-A1B | 79.9 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Qwen3.8-27B | 91.1 | % | Hugging Face model card | link | |
| diffusiongemma-26B-A4B-it | 0.3 | avg edit distance | Hugging Face model card | link | |
| gemma-4-12B-it | 0.2 | avg edit distance | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| Ling-3.0-flash | 73.6 | % | Hugging Face model card | link | |
| Ornith-1.5-35B-A3B | 67.8 | % | Hugging Face model card | link | |
| Ornith-1.5-9B | 59.5 | % | Hugging Face model card | link |
| Model | Score | Metric | Subtask | Source | URL |
|---|---|---|---|---|---|
| granite-4.2-30b | 62 | % | avg | Hugging Face model card | link |
| granite-4.2-8b | 58.1 | % | avg | Hugging Face model card | link |
| Spark-X2.5-4B | 30.4 | % | Hugging Face model card | link |