Benchmark
SkillsBench
Coding benchmarks compare models on programming tasks. Scores are percentages. Higher is better.
- Models and systems measured
- 27
- Scores recorded
- 32
- Sources
- 3
- IQ weight
- 1.99
Score distribution
- 54.4%
- Mean
- 55.3%
- Median
- 10.7%
- Std dev
- 44.3%
- 10th percentile
- 68.7%
- 90th percentile
- 28.7%
- Min
This benchmark has a weight of 1.99 for chat models, measured over 26 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model or system | Score |
|---|---|---|
| 1 | GPT-5.6 SolLing-3.1-flash official chart · SkillsBench · version, task count, harness, reasoning setting and run count not stated· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 73.5% |
| 2 | DeepSeek V4.1 FlashLing-3.1-flash official chart · SkillsBench · version, task count, harness, reasoning setting and run count not stated· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 71.3% |
| 3 | Qwen3.8 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 70.2% |
| 4 | Ling 3.1 Flash· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim | 68.7% |
| 5 | Ling 3.1 FlashLing-3.1-flash official chart · SkillsBench · version, task count, harness, reasoning setting and run count not stated· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 68.7% |
| 6 | GPT 5.5Dots Studio official appendix · unstarred comparison cell · version, task count, harness and reasoning setting not independently specified· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 66.5% |
| 7 | Claude Opus 5Ling-3.1-flash official chart · SkillsBench · version, task count, harness, reasoning setting and run count not stated· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 63.7% |
| 8 | Kimi K3Ling-3.1-flash official chart · SkillsBench · version, task count, harness, reasoning setting and run count not stated· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 63.6% |
| 9 | LongCat 2.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 63.3% |
| 10 | Hy4 preview· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 62.9% |
| 11 | GLM 5.3 FlashLing-3.1-flash official chart · SkillsBench · version, task count, harness, reasoning setting and run count not stated· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 61.3% |
| 12 | GLM 5.3Ling-3.1-flash official chart · SkillsBench · version, task count, harness, reasoning setting and run count not stated· Reported by Ant inclusionAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 59.4% |
| 13 | Qwen3.7 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 59.2% |
| 14 | GLM 5.1· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim | 58.4% |
| 15 | Hy3 (Deep Infra)· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 55.3% |
| 16 | Hy3 (Deep Infra)Dots Studio official appendix · unstarred comparison cell · version, task count, harness and reasoning setting not independently specified· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 55.3% |
| 17 | Qwen3.7 Plus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 54.9% |
| 18 | Kimi K2.6· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim | 54% |
| 19 | MiniMax M3· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim | 53% |
| 20 | DeepSeek V4 Pro· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim | 50.1% |
| 21 | Qwen3.6 27B· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 48.2% |
| 22 | Qwen3.6 Plus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 45.7% |
| 23 | Ling 3.0 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 44.8% |
| 24 | DeepSeek V4 Flash· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim | 44.7% |
| 25 | DeepSeek V4 FlashDots Studio official appendix · unstarred comparison cell · version, task count, harness and reasoning setting not independently specified· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 44.7% |
| 26 | Muse Glimmer 30B· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 44.3% |
| 27 | Muse Glimmer 30Bwith skills · high reasoning· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 44.3% |
| 28 | openPangu 2.0 FlashopenPangu official table · Thinking · Avg@5 · version, task subset, harness and detailed thinking budget not stated· Reported by Huawei openPangu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 42.6% |
| 29 | openPangu 2.0 FlashopenPangu official table · Non-Thinking · Avg@5 · version, task subset, harness and detailed thinking budget not stated· Reported by Huawei openPangu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 40% |
| 30 | MiniMax M2.7· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim | 34.9% |
| 31 | Qwen3.6 35B A3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 28.7% |
Not ranked
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | Score |
|---|---|
| DeepSeek V4 Pro Preview (SkillsBench table label)Dots Studio official appendix · unstarred comparison cell · version, task count, harness and reasoning setting not independently specified· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 50.1% |
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Hugging Face official benchmark leaderboards; licence varies by benchmark · Data by LLM Stats