PDEAgent-Bench: leaderboard
Metric: Case pass rate (%; 645 DOLFINx cases, single-shot, temperature 0; a case passes when the solver executes, stays within 10x the calibration error and 3x the calibration runtime). Source: arxiv.org. Saturation forecast: Around December 2026. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 54.1 |
| 2 | Claude Opus 4.7 | 47.8 |
| 3 | GPT-5.4 | 46 |
| 4 | Qwen 3.6 Plus | 23.4 |
| 5 | DeepSeek V3.2 | 4.2 |
Interactive version: theaggregate.ai/benchmark?slug=pdeagent-bench · How It Works · Data refreshed daily, snapshot 2026-09-25.