AutoBench: leaderboard
Automated multi-metric LLM evaluation ranking models by cost, latency, and domain performance. Models ranked by aggregate score across practical business tasks.
Metric: Judge Rating (1-5). Source: autobench.org. Status: saturation imminent. 38 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 Pro | 4.48 |
| 2 | GPT-5.2 | 4.43 |
| 3 | Gemini 3 Pro (Preview) | 4.41 |
| 4 | Claude Opus 4.5 | 4.39 |
| 5 | GPT-5.1 | 4.38 |
| 6 | Kimi K2 (Thinking) | 4.32 |
| 7 | Claude Sonnet 4.5 | 4.3 |
| 8 | Gemini 3 Flash (Preview) | 4.3 |
| 9 | Gemini 2.5 Pro | 4.29 |
| 10 | GPT-5 Mini | 4.29 |
| 11 | Grok 4.1 Fast (Reasoning) | 4.21 |
| 12 | Grok 4 | 4.2 |
| 13 | Qwen 3 235B A22B 2507 (Thinking) | 4.2 |
| 14 | GPT-OSS-120B | 4.18 |
| 15 | Gemini 2.5 Flash | 4.17 |
Interactive version: theaggregate.ai/benchmark?slug=autobench · How It Works · Data refreshed daily, snapshot 2026-09-05.