DeepSeek-V4-Flash vs Qwen2.5 32B Instruct
DeepSeek vs Alibaba — benchmarks, pricing, and capabilities side by side.
- •DeepSeek-V4-Flash has the higher intelligence index (89.4 vs 66.7)
- •Qwen2.5 32B Instruct is cheaper ($0.00 vs $0.10 per 1M input)
| DeepSeek-V4-Flash | Qwen2.5 32B Instruct | |
|---|---|---|
| Intelligence index | 89.4 | 66.7 |
| Developer | DeepSeek | Alibaba |
| Type | LLM | LLM |
| Access | Open weights | Open weights |
| Context window | 1,048,576 tokens | — |
| Input price | $0.10 / 1M | $0.00 / 1M |
| Output price | $0.20 / 1M | $0.00 / 1M |
| Speed | 109 tok/s | — |
| Released | April 24, 2026 | September 19, 2024 |
| Parameters | 284B (13B active) | 32500000000 |
| Input modalities | Text | — |
| Output modalities | Text | — |
Shared benchmarks
DeepSeek-V4-Flash
Qwen2.5 32B Instruct
GPQA Diamond
89.4
49.5
Humanity’s Last Exam
32.1
3.8