DeepSeek-V4 preview online, with million-token context as standard
Original title:DeepSeek-V4 预览版:迈入百万上下文普惠时代
AI summary
The DeepSeek-V4 preview is officially online and open-sourced, featuring a 1M ultra-long context, with agent capabilities, world knowledge and reasoning performance all reaching leading domestic and open-source levels. The model comes in V4-Pro and V4-Flash versions, and the API has been updated simultaneously, supporting OpenAI and Anthropic interfaces, with a maximum context of 1M and support for thinking mode and the reasoning_effort parameter.
Key Takeaways
- The DeepSeek-V4 preview is officially online and open-sourced,featuring a 1M ultra-long…
- The model comes in V4-Pro and V4-Flash versions,and the API has been updated…
- with a maximum context of 1M and support for thinking mode and the reasoning_effort parameter.
Quick Facts
- Published
- Updated
- Source
- 公众号:DeepSeek(深度求索)
- Category
- Model Releases
- AI score
- 85
百万上下文成为标配,长文档处理与多步 Agent 任务从实验走向生产,原先受限于输入长度的应用有了新的落地空间。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: DeepSeek-V4 preview online, with million-token context as standard. April 24, 2026. https://new.nat6.net/en/news/cmsdu3h2d015drofzvgy69u7x/