← Models
DeepSeek-V4-Flash-0731
DeepSeek · open weight · deepseek-v4-flash-0731
DeepSeek-V4-Flash-0731 is the official release of DeepSeek-V4-Flash, superseding the preview version with substantially stronger agentic capabilities. It retains the V4-Flash architecture and adds an attached DSpark speculative-decoding module for faster inference. The open-weight model supports a 1M-token context window and low, high, and max reasoning-effort levels, and DeepSeek reports that it outperforms the much larger DeepSeek-V4-Pro Preview across its published agent benchmark suite.
Benchmark scores
| Benchmark | Score |
|---|---|
| Terminal-Bench 2.1 | 0.83 |
| CyberGym | 0.77 |
| Toolathlon | 0.70 |
Pricing
- Novita$0.14 / $0.28
- DeepInfra$0.06 / $0.18
Input / output per 1M tokens
AA metrics
No Artificial Analysis link yet.
Arena Elo
- No Arena snapshot linked.