In the News: September 10, 2026 (Extra 2)
DeepSeek retires V4-Pro and moves coding-agent traffic to a cheaper V4.1-Flash model, with a hard cutover date of September 14.
Extra edition
Machine-readable
Download Markdown
Story
DeepSeek will route V4-Pro traffic to its cheaper V4.1-Flash model
DeepSeek released V4.1-Flash, a 552-billion-parameter model built on a new encoder-decoder architecture. DeepSeek says its key-value cache needs a quarter of the HBM and an eighth of the SSD storage used by the previous generation.
Read story →