DeepSeekLaunchAI ModelsJuly 31, 2026

DeepSeek releases V4-Flash-0731 open-weights model

DeepSeek released V4-Flash-0731, a 304B-parameter open-weights model scoring 50 on the Artificial Analysis Intelligence Index, a 10-point jump over the April V4 Flash. Priced at $0.14/M input and $0.27/M output tokens, it surpasses V4-Pro-Preview on agentic benchmarks and supports the Responses API.

How this story unfolded

3 weeks · 3 reports · 25 community posts · 28 of 30 shown

Jul 31
Aug 1
Aug 2
- Ran DS V4-Flash-0731 Locally on 3xMI50 32GB @ ~15 t/s TG
Aug 3
- DeepSeek V4-Flash (284B MoE) at 33 tok/s single / 68 tok/s aggregate on 2× RTX 3090 + a used quad-Xeon DDR4 server — full config
Aug 4
- [Deepseek-V4-Flash-0731] Full 1M context on a single RTX5090 + DDR5 Desktop Setup with VLLM CPU/Ram Offloading, ~800 tps pp & 15+ tps decode [Agentic Coding]
Aug 20
- The boring way to run Deepseek V4 Flash-0731 130-150 tks - 16x5060ti 16GB over 2 PLX88096 switches

DeepSeek by email

Get an email when DeepSeek has news

No news that day, no email.

DeepSeek releases V4-Flash-0731 open-weights model

How this story unfolded

DeepSeek by email

More stories today