Factorio Learning Environment - Lab Task Success Rate: leaderboard
Factorio Learning Environment (FLE) lab play: the percentage of structured tasks with fixed resources, in which an LLM agent must program a Factorio factory that reaches a target production of an item, that the agent completes. From the FLE leaderboard; higher is better, 0-100 (%).
Metric: Lab task success rate (%). Source: jackhopkins.github.io. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.5 Sonnet | 21.9 | #362 |
| 2 | GPT-4o | 16.6 | #357 |
| 3 | DeepSeek V3 | 15.1 | #329 |
| 4 | Gemini 2.0 Flash | 13 | #351 |
| 5 | Llama 3.3 70B | 5.2 | #578 |
| 6 | GPT-4o Mini | 4.2 | #598 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=factorio-learning-environment-lab-task-success-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.