Overview
DeepSeek V4 is a preview version of the new flagship MoE large model released by DeepSeek, featuring world-class reasoning performance and significantly enhanced Agent capabilities. It has been launched on the web, app, and API. The official website does not provide specific technical parameters, pricing strategies, financing events, or other detailed information. Relevant content in existing evaluations may be outdated or simplified.
Update (2026/9/14): DeepSeek announced that from 12:00 Beijing time on September 14 until V4.1 Pro ships, all requests to deepseek-v4-pro route to V4.1 Flash and bill at V4.1 Flash rates. DeepSeek says testing shows V4.1 Flash now surpasses V4 Pro across performance, cost, speed and total time, so V4 Pro is being wound down in order; users who still need it can contact DeepSeek to keep the service running.
Key Features
- World-Class Reasoning Performance: The preview version offers world-class reasoning performance, with significantly enhanced Agent capabilities.
- Multi-Platform Launch: It has been launched on the web, app, and API, supporting free conversations and quick integration.
- API Open Platform: Access the latest DeepSeek models for quick integration and a smooth experience.
- V4 Pro Traffic Routing: From 12:00 Beijing time on September 14 until V4.1 Pro ships, all requests to deepseek-v4-pro route to V4.1 Flash and bill at V4.1 Flash rates, with V4 Pro being wound down in order.
Use Cases
- Developers building applications that require top-tier reasoning capabilities
- Enterprise users seeking efficient Agent support
- Teams integrating DeepSeek models via the API open platform
- Users accessing DeepSeek V4 through the web, app, or API
Pros
- World-class reasoning performance, excels in handling complex tasks
- Significantly enhanced Agent capabilities, supporting smarter automated interactions
- Multi-platform availability (Web, App, API), flexible and free to use
- API open platform provides quick integration and smooth experience
- As a flagship MoE model, featuring advanced technical architecture
- Requests to deepseek-v4-pro route to V4.1 Flash and bill at V4.1 Flash rates from 2026/9/14
Pricing
From 12:00 Beijing time on September 14 until V4.1 Pro ships, all requests to deepseek-v4-pro route to V4.1 Flash and bill at V4.1 Flash rates.
Summary
DeepSeek V4 is suitable for developers and enterprise users who require top-tier reasoning capabilities and efficient Agent support. Its core advantages lie in world-class reasoning performance, free availability across multiple platforms, and convenient API integration. As of 2026/9/14, requests to deepseek-v4-pro are routed to V4.1 Flash and billed at V4.1 Flash rates while V4 Pro is wound down in order.
Version History
- All DeepSeek-V4-Pro traffic routed to V4.1-Flash (2026-09-14): DeepSeek announced that from 12:00 Beijing time on September 14 until V4.1 Pro ships, all requests to deepseek-v4-pro route to V4.1 Flash and bill at V4.1 Flash rates. DeepSeek says testing shows V4.1 Flash now surpasses V4 Pro across performance, cost, speed and total time, so V4 Pro is being wound down in order; users who still need it can contact DeepSeek to keep the service running.