FLY-EVAL++ - Single-Step Prediction - Predictive Accuracy: leaderboard
Metric: S1 D5 predictive score (%). Source: arxiv.org. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 92.8 |
| 2 | Gemini 3 Pro | 92.6 |
| 3 | O4 Mini | 92.5 |
| 4 | Kimi K2 (Thinking) | 92.3 |
| 5 | Kimi K2 0905 | 92.1 |
| 6 | DeepSeek R1 0528 | 92 |
| 7 | Llama 3.1 405B | 92 |
| 8 | GPT-4o | 91.9 |
| 9 | DeepSeek V3 | 91.8 |
| 10 | DeepSeek V3.1 | 91.8 |
| 11 | Claude 3.7 Sonnet (20250219) | 91.8 |
| 12 | Claude Sonnet 4.5 | 91.7 |
| 13 | Qwen 3 235B A22B | 91.7 |
| 14 | Qwen 2.5 32B | 91.7 |
| 15 | Seed-1.6 | 91.6 |
Interactive version: theaggregate.ai/benchmark?slug=fly-eval-plus-plus-single-step-prediction-predictive-accuracy · How It Works · Data refreshed daily, snapshot 2026-09-19.