Job Bench
Can the model produce useful work in a professional knowledge-work setting?
Knowledge Work9 ranked models11 reported values2 reportsHigher is better
Top rankings
One row per model, using its best reported score across effort settings.
| Rank | Model | Best score | Best reported setting |
|---|---|---|---|
| 1 | Claude Fable 5Anthropic | 57.4% | max effortMoonshot AI report1 value · 1 reportJul 17, 2026 · source |
| 2 | Muse Spark 1.1Meta | 54.7% | xhigh effortMeta report1 value · 1 reportJul 9, 2026 · source |
| 3 | Kimi K3Moonshot AI | 52.9% | max effortMoonshot AI report1 value · 1 reportJul 17, 2026 · source |
| 4 | Claude Opus 4.8Anthropic | 48.4% | max effortMoonshot AI report2 values · 2 reportsJul 17, 2026 · source |
| 5 | GPT-5.6 SolOpenAI | 46.5% | max effortMoonshot AI report1 value · 1 reportJul 17, 2026 · source |
| 6 | GLM-5.2Z.ai | 43.4% | max effortMoonshot AI report1 value · 1 reportJul 17, 2026 · source |
| 7 | GPT-5.5OpenAI | 38.3% | xhigh effortMoonshot AI report2 values · 2 reportsJul 17, 2026 · source |
| 8 | Muse Spark 1.0Meta | 17.0% | Reported configurationMeta report1 value · 1 reportJul 9, 2026 · source |
| 9 | Gemini 3.1 ProGoogle | 15.9% | high effortMeta report1 value · 1 reportJul 9, 2026 · source |
Effort curve
Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.
No cost-linked effort sweep for this version.
Definition and comparison boundarypublic methodology
Can the model produce useful work in a professional knowledge-work setting?
Professional analysis, document, finance, office, or domain-specific tasks. Higher is better. The value is the percentage reported in this lab's table.
Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.
65 professional tasks; OpenCode harness with Grok 4.3 judge; competitor values from the cited leaderboard.
Method / source