QuantCode-Bench - Backtest Pass: leaderboard
Metric: Backtest pass rate (%), single turn: share of the tasks whose generated strategy compiles and runs the backtest without runtime errors, on QuantCode-Bench (400 Backtrader trading-strategy tasks from Reddit, TradingView, StackExchange, GitHub and synthetic sources, multi-timeframe historical data); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 98.2 |
| 2 | GPT-5.4 | 95.5 |
| 3 | GPT-5.2 Codex | 94.5 |
| 4 | GLM-5 | 92.4 |
| 5 | Claude Sonnet 4.5 | 91.5 |
| 6 | Kimi K2.5 | 87.3 |
| 7 | Claude Sonnet 4.6 | 85.8 |
| 8 | Gemini 3 Flash | 76 |
| 9 | DeepSeek V3.2 | 75.8 |
| 10 | Qwen 3 235B A22B | 72.5 |
| 11 | Grok 4.1 Fast | 70.2 |
| 12 | Qwen 3 Coder 30B A3B Instruct | 59 |
| 13 | Gemini 2.5 Flash | 49.2 |
| 14 | Qwen 3 14B | 42.2 |
| 15 | Qwen 3 8B | 31.8 |
Interactive version: theaggregate.ai/benchmark?slug=quantcode-bench-backtest-pass · How It Works · Data refreshed daily, snapshot 2026-10-07.