BabyVision — Explicit python-tool mode, distinct tools axis from base baby_vision (no-tools)
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Claude Fable 5 | Anthropic | 90.5 | 1 | 2026-08-24 |
| 2 | GPT-5.6 Sol | OpenAI | 88.9 | 1 | 2026-08-24 |
| 3 | Kimi K3 | Moonshot | 85.7 | 1 | 2026-08-24 |
| 4 | Claude Opus 4.8 | Anthropic | 81.2 | 1 | 2026-08-24 |
| 5 | GPT-5.4 | OpenAI | 80.2 | 1 | 2026-05-03 |
| 6 | Kimi K2.6 | Moonshot | 68.5 | 1 | 2026-05-03 |
| 7 | Gemini 3.1 Pro | 68.3 | 2 | 2026-05-03 | |
| 8 | Kimi K2.5 | Moonshot | 40.5 | 1 | 2026-05-03 |
| 9 | Claude Opus 4.6 | Anthropic | 38.4 | 1 | 2026-05-03 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.