DeepSeek V4 preview is here! 1M context length is now standard, with an updated API.
DeepSeek V4 Preview Release Now Live and Open-Sourced
Original: DeepSeek V4 Preview Release | DeepSeek API Docs
Importance: 新モデルのリリースは多くのユーザーに影響を与えるため。
Summary
The DeepSeek V4 preview is officially live, introducing cost-effective 1M context length. It features DeepSeek-V4-Pro, rivaling top closed-source models, and DeepSeek-V4-Flash as a fast, economical option. Both are available now, with an updated API. DeepSeek enhances agent capabilities, excelling in reasoning for math and coding tasks.
Key Points
- DeepSeek V4 preview is officially released
- 1M context length is standardized
- DeepSeek-V4-Pro features 1.6T parameters
- DeepSeek-V4-Flash has 284B parameters
- API updated to support both models
View developer notes (APIs, breaking changes, migration)
The DeepSeek V4 preview is now live, standardizing 1M context length. DeepSeek-V4-Pro boasts 1.6T parameters, matching top models in performance, while DeepSeek-V4-Flash focuses on speed and cost with 284B parameters. The API is updated to support OpenAI ChatCompletions and Anthropic APIs.
Source: https://api-docs.deepseek.com/news/news260424
Outlet: DeepSeek Release Notes
This article is an AI-generated summary (OpenAI GPT-4o-mini) of publicly available information from Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, Sakana, and other vendors. The original source URL is always provided in accordance with fair-use citation requirements. Summaries are AI-generated and may contain mistranslations or misinterpretations. Always verify details with the original source.