BenchTable - Utility: leaderboard
Dubesor's small-scale personal benchmark, category score for the 11 prompt-adherence and utility tasks: difficulty-weighted points (pass, refine, fail, refusal) divided by the best points achieved on each task across models. Percent scale, higher is better; scores can be negative. The benchmark is archived, so the table no longer changes.
Metric: Weighted Score (%). Source: dubesor.de. Saturation forecast: Around August 2027. 347 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K2.5 (Thinking) | 87.4 | #152 (Kimi K2.5) |
| 2 | Grok 4.20 (Reasoning) | 87.4 | #179 (Grok 4.20) |
| 3 | Claude Opus 4.7 | 84.5 | #51 |
| 4 | GPT-5 Chat | 83.6 | #250 |
| 5 | GPT-4o | 82.6 | #357 |
| 6 | Grok 4 | 82.6 | #189 |
| 7 | Kimi K2 (Thinking) | 82.6 | #252 (Kimi K2) |
| 8 | DeepSeek V3.2 Speciale | 82.6 | #138 |
| 9 | Llama 3.1 405B Instruct | 81.5 | #470 |
| 10 | Claude Sonnet 4.6 | 80.1 | #94 |
| 11 | Gemini 3.1 Flash Lite (Preview) | 79.7 | #201 |
| 12 | GPT-4 Turbo | 79.5 | #417 |
| 13 | Gemini 2.5 Pro | 79 | #159 |
| 14 | Gemini 3 Flash (Preview) | 78.8 | #90 |
| 15 | Gemini 3 Flash (Preview) (Thinking) | 78.8 | #90 (Gemini 3 Flash (Preview)) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=benchtable-utility · How It Works · Data refreshed daily, snapshot 2026-11-05.