DeepSeek: DeepSeek V4 Flash 0731
DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows.
Everything else
| Max output | 384K |
|---|---|
| Input / 1M | $0.08 |
| Output / 1M | $0.18 |
| Cache read / 1M | $0.016 |
| Cache write / 1M | — |
| 1M in + 1M out | $0.26 |
| Takes | text |
| Returns | text |
| Tools | yes |
| Reasoning | optional · max, high, low |
| Open weights | yes |
| Released | 2026-07-31 |
| Knowledge cutoff | — |
| Retires | — |
| Catalog id | deepseek/deepseek-v4-flash-0731 |
Benchmarks
| source | epoch |
|---|---|
| eci | 152.53 |
| confidence | Likely |
| accessibility | Open weights (unrestricted) |
| gpqa | 0.91 |
| gpqaStderr | 0.0177 |
| aime | 0.944 |
| aimeStderr | 0.0209 |
| simpleQA | 0.347 |
| simpleQAStderr | 0.0151 |
Benchmark scores by Epoch AI, CC BY 4.0.
Using it
In 00, this model is picked per agent — and per tier, so the model that answers your customers need not be the one that writes your code. Get 00.