QuantCodeEval (OpenCode): leaderboard
QuantCodeEval — reproducing quantitative-finance strategies from 30 research papers, run under the OpenCode agent harness; the widest slice of the release, with every model on an identical scaffold. Full Pass Rate (%) counts an attempt only when every one of the 572 property checkers in the task contract passes.
Metric: Full Pass Rate (%). Source: www.quantcodeeval.cloud. Saturation forecast: Around December 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K3 | 26 | #35 |
| 2 | Claude Opus 4.7 | 22.67 | #51 |
| 3 | GPT-5.5 | 21.33 | #33 |
| 4 | Claude Sonnet 4.6 | 18.67 | #94 |
| 5 | Qwen 3.7 Max | 18 | #85 |
| 6 | MiniMax-M3 | 14.67 | #151 |
| 7 | GLM-5.2 | 14 | #80 |
| 8 | DeepSeek V4 Pro | 12 | #112 |
| 9 | GLM-5.1 | 12 | #109 |
| 10 | Qwen 3.6 Plus | 10.67 | #124 |
| 11 | MiniMax-M2.7 | 3.33 | #270 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=quantcodeeval-opencode · How It Works · Data refreshed daily, snapshot 2026-11-05.