DeepSeek-R1-Lite-Preview: benchmark results
DeepSeek's first reasoning preview, offered only in web chat; its promised API and open weights shipped as DeepSeek-R1 instead. Provider: DeepSeek. Released 2024-11-20. Access: Interface.
Aggregate Rating 1572 ± 24, rank #1373 of 10916 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SuperCLUE-Math6o (November 2024) - Overall | 64.66 | Score | 97 |
| SuperCLUE-Math6o - Overall | 69.87 | Score | 91.4 |
| SuperCLUE General (December 2024) - Science | 69.7 | Score | 81.1 |
| SuperCLUE General (December 2024) - Overall | 63.8 | Score | 67.6 |
| SuperCLUE-CoT (October 2024) - Overall | 45.8 | Score | 65.7 |
| SuperCLUE General (December 2024) - Hard | 44.9 | Score | 60.8 |
| SuperCLUE General (December 2024) - Liberal Arts | 76.8 | Score | 56.8 |
| BenchTable - Reasoning | 42.4 | Weighted Score (%) | 54.9 |
| SuperCLUE-MathCLUE - Overall | 61.19 | Score | 40.7 |
| BenchTable | 42.4 | Total Score (%) | 40.4 |
| BenchTable - Tech | 45.2 | Weighted Score (%) | 40.4 |
| BenchTable - STEM | 39.5 | Weighted Score (%) | 35.2 |
Interactive version: theaggregate.ai/model?slug=deepseek-r1-lite-preview · How It Works · Data refreshed daily, snapshot 2026-11-08.