REAL Evals: leaderboard
Realistic Evaluations for Agents in the Loop: real-world task completion benchmark with verified model performance on practical tasks.
Metric: Task Completion (%). Source: www.realevals.xyz. Saturation forecast: Around January 2027. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Flash | 30 | #245 |
| 2 | Claude Sonnet 4.5 | 29.2 | #142 |
| 3 | Gemini 2.5 Pro | 25.3 | #152 |
| 4 | Claude Sonnet 4 | 24 | #199 |
| 5 | Claude 3.7 Sonnet (Thinking) | 24 | #249 (Claude 3.7 Sonnet) |
| 6 | GPT-5 | 23.6 | #98 |
| 7 | Claude Opus 4 (Thinking) | 19.7 | #163 (Claude Opus 4) |
| 8 | Grok 4 Fast | 19.3 | #254 |
| 9 | Claude Sonnet 4 (Thinking) | 18.9 | #199 (Claude Sonnet 4) |
| 10 | O3 | 16.7 | #124 |
| 11 | Claude 3.7 Sonnet | 16.3 | #249 |
| 12 | DeepSeek V3.2 Exp | 12.4 | #234 |
| 13 | O3 Mini | 12 | #272 |
| 14 | GPT-4.1 | 11.6 | #248 |
| 15 | GPT-5 Nano | 9.9 | #425 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=real-evals · How It Works · Data refreshed daily, snapshot 2026-10-20.