Taste-Bench - Mean Presentation Accuracy: leaderboard
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around March 2028. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol (xHigh) | 65.2 | #18 (GPT-5.6 Sol) |
| 2 | GPT-5.5 (xHigh) | 64.6 | #29 (GPT-5.5) |
| 3 | GLM-5.2 (Max) | 64 | #70 (GLM-5.2) |
| 4 | GPT-5.6 Terra | 62.2 | #40 |
| 5 | Grok 4.5 (High) | 62 | #46 (Grok 4.5) |
| 6 | Claude Sonnet 5 (Thinking) | 60.7 | #56 (Claude Sonnet 5) |
| 7 | Claude Opus 5 | 60.3 | #11 |
| 8 | GPT-5.6 Luna | 58.9 | #67 |
| 9 | MiniMax-M3 | 57.2 | #130 |
| 10 | DeepSeek V4 Flash (Max) | 54.9 | #134 (DeepSeek V4 Flash) |
| 11 | Mistral Medium 3.5 (High) | 51.8 | #338 (Mistral Medium 3.5) |
| 12 | GPT-5.4 Mini (High) | 49.5 | #206 (GPT-5.4 Mini) |
| 13 | GPT-5.4 Nano (High) | 48.8 | #320 (GPT-5.4 Nano) |
| 14 | Grok 4.20 (Reasoning) | 29 | #169 (Grok 4.20) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=taste-bench-mean-presentation-accuracy · How It Works · Data refreshed daily, snapshot 2026-10-13.