{"id":4626,"date":"2026-09-26T18:16:29","date_gmt":"2026-09-26T09:16:29","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/2026\/09\/26\/deepseek-v4-pro-0813-released\/"},"modified":"2026-09-27T20:08:46","modified_gmt":"2026-09-27T11:08:46","slug":"deepseek-v4-pro-0813-released","status":"publish","type":"post","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/26\/deepseek-v4-pro-0813-released\/","title":{"rendered":"DeepSeek-V4-Pro-0813 Text Generation Model: ~998GB Memory, GGUF Builds"},"content":{"rendered":"

<\/p>\n

At a Glance<\/h2>\n
\n\n\n\n\n\n\n\n\n\n\n\n\n
Item<\/th>\nValue<\/th>\n<\/tr>\n<\/thead>\n
Repository<\/td>\ndeepseek-ai\/DeepSeek-V4-Pro-0813<\/a><\/td>\n<\/tr>\n
Family guide<\/td>\nDeepSeek-V4 guide (2 articles)<\/a><\/td>\n<\/tr>\n
Publisher guide<\/td>\nDeepSeek: models and licenses<\/a><\/td>\n<\/tr>\n
Published<\/td>\n2026-08-13<\/td>\n<\/tr>\n
License<\/td>\nmit<\/td>\n<\/tr>\n
Formats<\/td>\nsafetensors<\/a><\/td>\n<\/tr>\n
Paper<\/td>\narXiv:2606.19348<\/a><\/td>\n<\/tr>\n
Source type<\/td>\nPrimary source (the publisher itself)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

Values determined by this site’s code at collection time. Dates are JST.<\/em><\/p>\n

<\/p>\n

Overview<\/h2>\n

DeepSeek has released the official version of its large-scale MoE model, DeepSeek-V4-Pro-0813<\/strong>, on Hugging Face. According to the model card, it replaces the previous preview version (DeepSeek-V4-Pro (Preview)) with significantly enhanced agent capabilities, bringing noticeable performance improvements especially in production environments. The structure remains the same as the preview version, with the addition of the DSpark<\/strong> module for speculative decoding.<\/p>\n

According to the summary of the technical report for the DeepSeek-V4 series, the series is designed to be “a highly efficient model capable of routinely handling 1-million-token contexts.” At a 1-million-token context, it is reported to require 27% of the inference compute (FLOPs) per token and 10% of the KV cache compared to DeepSeek-V3.2.<\/p>\n

Specifications<\/h2>\n