DeepSeek R1 Distill Qwen 32B
Zero-eval
by DeepSeek
+
+
+
+
About
DeepSeek-R1-Distill-Qwen-32B was created as a larger distilled variant, designed to transfer more of DeepSeek-R1's reasoning capabilities into a Qwen-based foundation. Built to serve applications requiring enhanced analytical depth, it represents a powerful option in the distilled reasoning model family.
+
+
+
+
Pricing Range
Input (per 1M)$0.12 -$0.12
Output (per 1M)$0.18 -$0.18
Providers1
+
+
+
+
Timeline
AnnouncedJan 20, 2025
ReleasedJan 20, 2025
+
+
+
+
Specifications
Training Tokens14.8T
+
+
+
+
License & Family
License
MIT
Performance Overview
Performance metrics and category breakdown
Overall Performance
4 benchmarks
Average Score
74.2%
Best Score
94.3%
High Performers (80%+)
2Performance Metrics
Max Context Window
256.0KAvg Throughput
37.0 tok/sAvg Latency
1ms+
+
+
+
All Benchmark Results for DeepSeek R1 Distill Qwen 32B
Complete list of benchmark scores with detailed information
| MATH-500 | text | 0.94 | 94.3% | Self-reported | |
| AIME 2024 | text | 0.83 | 83.3% | Self-reported | |
| GPQA | text | 0.62 | 62.1% | Self-reported | |
| LiveCodeBench | text | 0.57 | 57.2% | Self-reported |
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+