Revision History

Auto-saved before update

Auto-saved before update

Added Vision-Exp weights, architecture, and deployment details

Added Artificial Analysis's Aug. 24 hosted max-effort evaluation with text-oriented scope and route-specific caveats.

Added the experimental Vision-Exp API variant, image-input limits, vendor evaluation caveats, and current pricing and availability.

Correction: replaced two references to a source on the site's excluded list with IT之家 and Jiemian News coverage; the V4-Pro price reduction began as a late-April 2026 limited-time promotion, not a May permanent cut, and the peak-pricing plan was communicated in a June 29, 2026 email to users

Add Bloomberg Aug 4 'death zone' coverage to Reception; update OpenRouter listing to the separate 0731 entry with current routed pricing

Added Nous Portal/Novita AI availability of V4 Flash to the third-party redistribution section

Correction: three of the four repository commits are attributed to msr2000 alone

Correction: the section reconciling the 304B parameter figure ended by attributing it to the bundled DSpark module, contradicting the evidence set out immediately above it; the figure is a tensor-element accounting artifact

Open weights shipped: added the July 31 2026 MIT release of deepseek-ai/DeepSeek-V4-Flash-0731 on Hugging Face (48 shards, ~167 GB, DSpark speculative decoding module attached), with commit timeline, config.json specifications, vLLM/SGLang/local inference guidance and third-party quantisations. Added a Reading the parameter count subsection reconciling the 284B/13B figure against the 304B and 165B Hugging Face repository totals. Updated V4-Pro and Responses API status to August 1 2026, and corrected two benchmark row labels against the model card.

No summary provided