DeepSeek V4 Flash 0731 Open Weights: 284B Params Hits Top 3 Open Models
ArtificialAnlys · x · 2026-08-01
DeepSeek has released the weights for its new model, DeepSeek V4 Flash 0731. Scoring 50 on the Artificial Analysis Intelligence Index, it ranks among the top 3 open weights models globally. The weights are available under the MIT license, allowing for unrestricted commercial use and modification.
Architecturally, V4 Flash 0731 shares the same design and pricing as the earlier V4 Flash. With 284B total parameters (13B active) and mixed FP4/FP8 precision, the file size is approximately 167GB. It delivers a significant leap in intelligence for its size class, sitting on the Pareto frontier for Intelligence Index vs. Total Parameters. The model is also accessible via DeepSeek's first-party API.
Related event: DeepSeek V4-Flash Released and Open-Sourced with Flagship Performance(90 posts)→
More from Models
- Kimi Tops Fullstack Code Arena, Beating Anthropic and OpenAI Models — airesearch12 · 2026-08-01
- Cerebras Shares Guide to GPT-5.6: Sol, Terra, and Luna Models — iruletheworldmo · 2026-08-01
- DeepSeek v4 Flash Reported to Loop and Forget Context in Coding — kwizzle · 2026-08-01
- Local Deployment on DGX Spark: Exploring Upgrades Beyond Qwen 3.5 122B — Voxandr · 2026-08-01
- 25.5 Trillion Tokens a Day: How SuperPods Power Massive RL Training — zephyr_z9 · 2026-08-01
- Benchmarking Agent Harnesses: Kimi K3 Shines, Claude Code Costs 4x More — omarsar0 · 2026-08-01