Head to head
DeepSeek V4 Flash vs Qwen2.5 7B Instruct Turbo
On a blended workload of one million input tokens and three million output tokens, DeepSeek V4 Flash is 1.2× cheaper than Qwen2.5 7B Instruct Turbo.
| DeepSeek V4 Flash | Qwen2.5 7B Instruct Turbo | |
|---|---|---|
| Provider | DeepSeek | Together AI |
| Input / MTok | $0.14 | $0.3 |
| Output / MTok | $0.28 | $0.3 |
| Cache read / MTok | $0.0028 | — |
| Context window | 1,000,000 | 32,768 |
| Blended 1M in + 3M out | $0.98 | $1.20 |
| Status | ga | ga |