DeepSeek releases V4-Pro (1.6T/49B) and V4-Flash (284B/13B), MIT, 1M context

Made by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made

DeepSeek V4-Pro (1.6T/49B) and V4-Flash (284B/13B), MIT, 1M context

Why it matters

Re-entered the frontier while keeping MIT

Source — Primary sources first; secondary reports are labelled “Press”

Officialhuggingface.cohttps://huggingface.co/deepseek-ai/DeepSeek-V4-Pro

Added in Week 35, 2026 · See that week

On the map — ⑨ Open-Weight Lineage line

  1. Previous stopKimi K2 Thinking
  2. This recordDeepSeek releases V4-Pro (1.6T/49B) and V4-Flash (284B/13B), MIT, 1M context
  3. Next stopK-EXAONE 2.0
See it on the map

In the model family tree — DeepSeek family · 12 versions

  1. Earlier versionDeepSeek-V3.2
  2. This recordDeepSeek V4-Pro / V4-Flash
  3. Later versionDeepSeek-V4.1-Flash
See the whole DeepSeek family

Same week

DeepSeek-V4PaperCohere announces acquisition of and merger with Germany's Aleph AlphaPressGPT-5.5OfficialGoogle Cloud previews its 8th-generation TPUs at Cloud NextPress

Nearby on the same axis

Alibaba releases Qwen3.6-27B under Apache 2.0PressMoonshot AI releases Kimi K2.6PressLG AI Research releases EXAONE 4.5 33BOfficialMeta Superintelligence Labs releases Muse SparkOfficial