QwQ-32B-Preview
Zero-eval
by Alibaba Cloud / Qwen Team
+
+
+
+
About
QwQ 32B Preview was introduced as an early access version of the QwQ reasoning model, designed to allow researchers and developers to experiment with advanced analytical capabilities. Built to gather feedback on reasoning-enhanced architecture, it represents an experimental step toward more thoughtful language models.
+
+
+
+
Pricing Range
Input (per 1M)$0.15 -$1.20
Output (per 1M)$0.20 -$1.20
Providers4
+
+
+
+
Timeline
AnnouncedNov 28, 2024
ReleasedNov 28, 2024
Knowledge CutoffNov 28, 2024
+
+
+
+
License & Family
License
Apache 2.0
Base ModelQwen2.5 32B Instruct
Performance Overview
Performance metrics and category breakdown
Overall Performance
4 benchmarks
Average Score
63.9%
Best Score
90.6%
High Performers (80%+)
1Performance Metrics
Max Context Window
65.5KAvg Throughput
67.3 tok/sAvg Latency
1ms+
+
+
+
All Benchmark Results for QwQ-32B-Preview
Complete list of benchmark scores with detailed information
| MATH-500 | text | 0.91 | 90.6% | Self-reported | |
| GPQA | text | 0.65 | 65.2% | Self-reported | |
| AIME 2024 | text | 0.50 | 50.0% | Self-reported | |
| LiveCodeBench | text | 0.50 | 50.0% | Self-reported |
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+