DeepSeek V4 Flash Surfaces on OpenRouter: 284B Total Params, 1M Context
MikePFrank · x · 2026-08-05
OpenRouter page information indicates that the DeepSeek V4 Flash 0731 model has seemingly appeared. The model uses a sparse mixture-of-experts architecture with 284B total parameters and 13B active parameters. It features a 1M context window, with a listed release date of July 31, 2026.
Pricing is highly competitive at $0.09 per 1M tokens for input and $0.18 per 1M tokens for output. The official description notes the model is suited for coding, reasoning, and agent workflows.
More from Models
- OpenAI's Rumored 'Mewfour' Model Reportedly Enters Testing — thesaraharminta · 2026-08-05
- Frontier AI Model Hacks Company Autonomously for the 3rd Time in 2 Weeks — Miles_Brundage · 2026-08-05
- Image-to-WebDev Arena: Claude Opus 5 Max Takes the Lead — arena · 2026-08-05
- Deepgrove Releases Maple-Preview: 20B Ternary-Weight Open-Weight LLM — cafedude · 2026-08-05
- Deep Dive: Where Does DS4 Flash 0731 Sit for Coding Among Frontier Models? — Imjustmisunderstood · 2026-08-05
- Qwen Devs Tease Upcoming 27B Model with 'New Level of Capability' — cedric_chee · 2026-08-05