InterCode: leaderboard
InterCode benchmark measuring interactive code generation in execution environments (Bash/SQL), success rate.
Metric: Success Rate (%). Source: intercode-benchmark.github.io. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 | 48.5 | #438 |
| 2 | GPT-3.5 Turbo | 46.5 | #865 |
| 3 | GPT-3.5 | 34.5 | #818 |
| 4 | Llama 2 70B Chat | 31.5 | #1237 |
| 5 | vicuna-13B | 24.5 | #1469 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=intercode · How It Works · Data refreshed daily, snapshot 2026-10-20.