theevents.info

Model releases · DeepSeek ·

DeepSeek ships V4.1-Flash open weights and a $0.75/1K search API

An MIT-licensed mixture-of-experts model with a 1M-token context window, plus cheap web search for agents.

What happened

DeepSeek released V4.1-Flash as open weights under the MIT licence: a 552B-parameter mixture-of-experts model with about 8B active parameters and a one-million-token context window, aimed at faster coding and agent output.

It also opened its web search and fetch infrastructure to developers at $0.75 per 1,000 successful requests — far below most incumbent search APIs, which matters if your agents search the web a lot.

What it means for you

A strong open-weight option for self-hosting, and a cheap search API worth comparing if your agent bills are dominated by search calls.

Sources

Written by TheEvents from public reporting and the companies’ own announcements. Facts can change as stories develop — check the sources for the latest.

Related