DeepSeek promises its new AI model has 'world-class' reasoning

Source: Engadget

Overview

DeepSeek has released its latest AI models, the V4 Pro and V4 Flash, a little over a year after the company went viral and became the top‑rated free app on Apple’s App Store in the United States. In its announcement, DeepSeek declared, “Welcome to the era of cost‑effective 1 million context length.”

Context length refers to the maximum number of tokens an AI model can retain in a single conversation. A larger context window enables more coherent and consistent extended dialogues. For comparison, OpenAI’s recently announced GPT‑5.5 offers a context window ranging from 400 k to 1 M tokens.

Model Details

• DeepSeek‑V4‑Pro

  • Total parameters: 1.6 trillion
  • Active parameters: 49 billion
  • Enhanced agentic capabilities
  • Claims to rival top closed‑source models in reasoning and to trail only Gemini‑3.1‑Pro in world knowledge

• DeepSeek‑V4‑Flash

  • Total parameters: 284 billion
  • Active parameters: 13 billion
  • Faster response times than V4 Pro
  • Reasoning abilities close to V4 Pro and comparable performance on simple agent tasks

Both models remain open‑source, allowing users to download and modify the code as desired.

Controversy

Shortly after DeepSeek topped the App Store charts, it was banned for use by U.S. federal agencies and on government‑owned devices. Authorities cited national‑security concerns and a perceived threat to U.S. AI stocks. South Korea also paused downloads of the app over privacy worries.

Announcement

🚀 DeepSeek‑V4 Preview is officially live & open‑sourced! Welcome to the era of cost‑effective 1M context length.
🔹 DeepSeek‑V4‑Pro: 1.6T total / 49B active params. Performance rivaling the world’s top closed‑source models.
🔹 DeepSeek‑V4‑Flash: 284B total / 13B active params. … pic.twitter.com/n1AgwMIymu
— DeepSeek (@deepseek_ai) April 24, 2026