OPT-BENCH: leaderboard
Large Language Models (LLMs) have demonstrated remarkable capabilities in reasoning and tool use. However, the fundamental cognitive faculties essential for problem solving, including perception, reasoning, and memory, remain the stable core of intelligence. Unlike memorizing specific patterns, humans succeed in novel environments by applying these intrinsic faculties to adapt and optimize. Yet, w
Metric: EG (20 steps) (self-reported). Source: benchmarklist.com. Saturation forecast: Around October 2026. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 3 | 63 | #304 |
| 2 | Qwen 3 235B A22B (Thinking) | 62 | #313 (Qwen 3 235B A22B) |
| 3 | GPT-4o (2024-08-06) | 61 | #336 |
| 4 | DeepSeek V3.1 (Thinking) | 60 | #264 (DeepSeek V3.1) |
| 5 | GPT-4.1 | 58 | #248 |
| 6 | Qwen 3 32B | 58 | #443 |
| 7 | Qwen 3 30B A3B | 54 | #493 |
| 8 | Gemini 2.0 Flash | 53 | #342 |
| 9 | Claude 3.7 Sonnet (20250219) | 53 | #201 |
| 10 | Claude 3.5 Sonnet (20241022) | 48 | #291 |
| 11 | Qwen 2.5 72B Instruct | 45 | #454 |
| 12 | Qwen 2.5 32B Instruct | 45 | #498 |
| 13 | Qwen 3 8B | 38 | #674 |
| 14 | Qwen 2.5 14B Instruct | 30 | #647 |
| 15 | InternLM3-8B-Instruct | 30 | #867 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opt-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.