Benchmarks / Agentic
JobBench
Uses the first-party weighted-score leaderboard. Score: weighted score. Collected by Jake Boggs' benchmark dashboard.
JobBench
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5 | max | 65.7% |
| 2 | Muse Spark 1.3 | max | 64.9% |
| 3 | Muse Spark 1.2 | max | 61.6% |
| 4 | GLM 5.3 | max | 61.4% |
| 5 | Fable 5 | max | 57.4% |
| 6 | Muse Spark 1.1 | max | 54.7% |
| 7 | Kimi K3 | max | 54.3% |
| 8 | Qwen 3.8 Max | max | 52.7% |
| 9 | Opus 4.8 | max | 48.4% |
| 10 | GPT-5.6 Sol | max | 45.4% |
| 11 | Opus 4.7 | max | 44.5% |
| 12 | GLM 5.2 | max | 43.4% |
| 13 | GPT-5.5 | xhigh | 38.3% |
| 14 | Sonnet 4.6 | max | 36.6% |
| 15 | GPT-5.4 | xhigh | 32.2% |
| 16 | Gemini 3.5 Flash | xhigh | 31.5% |
| 17 | Inkling | max | 31.1% |
| 18 | Nemotron 3 Ultra 550B A55B | max | 27.2% |
| 19 | GPT-5.2 | xhigh | 26.6% |
| 20 | Sonnet 4.5 | high | 20.7% |
| 21 | Gemini 3.1 Pro | high | 15.9% |