LongProc — leaderboard
Long-context procedural generation benchmark testing models on generating long, structured outputs following complex procedural instructions.
Metric: Accuracy @ 0.5K (%). Source: princeton-pli.github.io. Status: saturated. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o (2024-08-06) | 94.8 |
| 2 | Gemini 1.5 Pro (001) | 89.2 |
| 3 | Gemini 1.5 Flash (001) | 78.9 |
| 4 | Claude 3.5 Sonnet (20241022) | 78.4 |
| 5 | Llama 3.1 70B Instruct | 72.9 |
| 6 | Qwen 2.5 72B Instruct | 68.7 |
| 7 | Llama 3.1 8B Instruct | 26.9 |
| 8 | Llama 3.2 3B Instruct | 13.5 |
| 9 | Llama 3.2 1B Instruct | 4 |
Interactive version: theaggregate.ai/benchmark?slug=longproc · How the rankings work · Data refreshed daily, snapshot 2026-07-22.