Benchmarks · agentic
ApprenticeBench CUA (NeoCognition)
A computer-use agent operates the Odoo ERP through its screens across 100 sequential accounts-payable tasks with diminishing mentoring. Maintained by NeoCognition. Score: Cumulative success rate.
18 models · 29 results · data as of 2026-10-29 · apprenticebench.com ↗
Leaderboard
Fable 5.1 max
All results
29 rows| # | Model | Effort | Score | Harness |
|---|---|---|---|---|
| 1 | Fable 5.1 | max | 72.0% | Claude Code |
| 2 | GPT-6 Astra Codex | max | 68.0% | Codex |
| 3 | GPT-6 Astra Codex | xhigh | 61.0% | Codex |
| 4 | Fable 5.1 | xhigh | 59.0% | Claude Code |
| 5 | Fable 5 | xhigh | 43.0% | Claude Code |
| 6 | Opus 5 | max | 36.0% | Claude Code |
| 7 | Fable 5 | high | 35.0% | Claude Code |
| 8 | Fable 5 | max | 34.0% | Claude Code |
| 9 | GPT-5.6 Sol Codex | max | 26.0% | Codex |
| 10 | Gemini 3.8 Flash | max | 24.0% | Claude Code |
| 11 | Fable 5 | medium | 23.0% | Claude Code |
| 12 | GPT-5.5 Codex | xhigh | 20.0% | Codex |
| 13 | GPT-5.6 Sol Codex | high | 20.0% | Codex |
| 14 | Fable 5 | low | 19.0% | Claude Code |
| 15 | GPT-5.6 Sol Codex | xhigh | 19.0% | Codex |
| 16 | Muse Spark 1.3 | max | 19.0% | Claude Code |
| 17 | Kimi K3 | max | 18.0% | Claude Code |
| 18 | Sonnet 5 | max | 16.0% | Claude Code |
| 19 | Gemini 3.7 Flash | max | 16.0% | Claude Code |
| 20 | GPT-5.6 Terra Codex | max | 16.0% | Codex |
| 21 | Grok 4.6 | max | 13.0% | Claude Code |
| 22 | GPT-5.4 Codex | xhigh | 11.0% | Codex |
| 23 | GPT-5.6 Sol Codex | medium | 10.0% | Codex |
| 24 | Opus 4.7 | max | 7.0% | Claude Code |
| 25 | GPT-5.6 Luna Codex | max | 7.0% | Codex |
| 26 | Opus 4.6 | max | 5.0% | Claude Code |
| 27 | GPT-5.6 Sol Codex | low | 4.0% | Codex |
| 28 | Sonnet 4.6 | max | 2.0% | Claude Code |
| 29 | GPT-5.6 Sol Codex | none | 0.0% | Codex |