DeepSeek
(Hangzhou, China)
Open-source frontier champion behind the DeepSeek-V4.1 family (552B Causal Encoder-Decoder MoE, native multimodal, MIT-licensed) with DSpark speculative decoding and 1M context.
What is the latest AI model from DeepSeek?
DeepSeek's primary flagship model is DeepSeek V4.1 Flash (552B MoE (8B / 16B Activated), 1,000,000 tokens context). Released Sept 10, 2026; first of DeepSeek's Causal Encoder-Decoder family with native image+text, 1M context and 384K output. Replaces V4-Flash and V4-Pro from Sept 14. How we define “latest”
Release Timeline
NEWEST FIRSTAugust 2026[2 releases]
Open-sourced under MIT license on August 30, 2026. 305B parameter multimodal vision-language model with native document understanding.
DeepSeek's primary 1.6T parameter powerhouse with 49B activated per token, configurable thinking budget, and 1M context.