DeepSeek V4
Open-weight MoE flagship models with 1M context and frontier-class reasoning
No preview available
About DeepSeek V4
Overview
DeepSeek V4 is a family of two open-weight Mixture-of-Experts language models released in April 2026: V4-Pro (1.6T parameters, 49B active) and V4-Flash (284B parameters, 13B active). Both support a 1 million token context window, dual Thinking/Non-Thinking modes, and OpenAI-compatible API access. DeepSeek claims V4-Pro leads open models on math, STEM, and agentic coding benchmarks while rivaling top closed-source frontier models.
Key Features
- Largest open-weight model available (V4-Pro at 1.6T parameters).
- Extremely competitive pricing, especially V4-Flash at $0.14/M input tokens.
- 1M token context window handles large codebases and long documents natively.
- Open-weight release allows self-hosting for privacy-sensitive workloads.
- Dual Thinking/Non-Thinking modes add flexibility for latency vs. accuracy trade-offs.
Build and discover AI agents for personal intelligence
The leading no-code platform for building web apps with AI.
An AI design agent for automated graphic design.
Meta’s general-purpose AI assistant for search, creation, and everyday help
The viral AI platform for chatting with any character.
AI-powered platform for creating and enhancing digital content