DeepSeek V4.1 Flash reaches Vercel AI Gateway: 1M context, one key
The fast open-weight model is one key away with spend tracking; Vercel also gives any model web search and fetch, and ships cron Schedules in beta.

Copy markdown
A cheap 1M-context model, one key away
Vercel AI Gateway added DeepSeek V4.1 Flash 1017 on Nov 7: a 1M-token-context open-weight model listed around $0.30/$1.20 per 1M tokens, callable with one key plus spend tracking and request traces — no separate DeepSeek account. Builders are already shipping with it; one reportedly built a Mac app for under $1.
Web search and fetch for any model
Vercel's new DeepSeek Search and Fetch tools (Nov 6) let any tool-calling model search the web and pull page content through one API key. You can give an agent live browsing without wiring up a separate search provider or scraper.
Cron Schedules land in beta
Vercel Schedules (public beta, Nov 5) runs tasks on a cron or at a single future time, dispatching to a Function or Queue topic in any environment — and agents can set dynamic schedules. Recurring agent jobs without standing up your own scheduler.