Benchmarks / Agentic
ApprenticeBench API (NeoCognition)
An agent completes the same 100 sequential accounts-payable tasks in the Odoo ERP with diminishing mentoring, using dedicated application API calls instead of the screen interface. Maintained by NeoCognition. Score: Cumulative success rate.
ApprenticeBench API (NeoCognition)
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Fable 5.1 | max | 70.0% |
| 2 | GPT-6 Astra Codex | max | 65.0% |
| 3 | Opus 5 | max | 49.0% |
| 4 | Fable 5 | max | 45.0% |
| 5 | Grok 4.6 | max | 45.0% |
| 6 | Grok 4.5 | max | 38.0% |
| 7 | Gemini 3.8 Flash | high | 34.0% |
| 8 | Gemini 3.7 Flash | high | 32.0% |
| 9 | GPT-5.6 Sol Codex | max | 30.0% |
| 10 | Opus 4.8 | max | 28.0% |
| 11 | GPT-5.5 Codex | xhigh | 27.0% |
| 12 | Sonnet 5 | max | 26.0% |
| 13 | Kimi K3 | max | 25.0% |
| 14 | Qwen 3.8 Max | max | 24.0% |
| 15 | Muse Spark 1.3 | max | 24.0% |
| 16 | GPT-5.6 Terra Codex | max | 21.0% |
| 17 | DeepSeek V4 Pro | max | 21.0% |
| 18 | GLM 5.3 Flash | max | 21.0% |
| 19 | Muse Spark 1.2 | max | 19.0% |
| 20 | DeepSeek V4 Flash | max | 18.0% |
| 21 | GLM 5.2 | max | 17.0% |
| 22 | Opus 4.7 | max | 14.0% |
| 23 | GPT-5.6 Luna Codex | max | 13.0% |
| 24 | Gemini 3.6 Flash | high | 12.0% |
| 25 | Muse Spark 1.1 | max | 12.0% |
| 26 | Kimi K2.6 | max | 10.0% |
| 27 | Qwen 3.8 Flash | max | 10.0% |