Long Code Arena: leaderboard
JetBrains long-context code benchmark: 5 tasks (library code generation, CI repair, commit message generation, bug localization, module summarization) testing real-world repository-scale understanding.
Metric: Mean Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 | 95.7 | #183 |
| 2 | Claude 3.5 Sonnet | 83.8 | #350 |
| 3 | DeepSeek R1 | 79.6 | #251 |
| 4 | GPT-4o | 70.2 | #343 |
| 5 | Gemini 1.5 Pro | 58.1 | #445 |
| 6 | Llama 3.1 405B | 46.8 | #466 |
| 7 | Claude 3 Haiku | 42.3 | #803 |
| 8 | Llama 3.1 70B | 28.8 | #592 |
| 9 | Llama 3.1 8B | 11.7 | #1151 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=long-code-arena · How It Works · Data refreshed daily, snapshot 2026-10-20.