Deepseek V4 on AI Gateway
DeepSeek V4 now ships on Vercel AI Gateway. Two variants. 1M token context. Ready for agentic workflows.
Why it matters
DeepSeek V4's availability on Vercel's unified API layer expands distribution to developers building agents and reasoning-heavy applications. This is infrastructure-level adoption, not just model release—it signals DeepSeek moving from closed research to mainstream developer tooling.
The key facts
7 to knowDeepSeek V4 Pro and V4 Flash variants available
1M token context window default across both models
V4 Pro optimized for agentic coding, formal mathematical reasoning, long-horizon workflows
V4 Flash positioned for high-volume, latency-sensitive workloads at lower cost
Integrated into Vercel AI Gateway with unified API, usage tracking, cost monitoring, retries, failover
Tool use support across MCP workflows and agent frameworks
Published April 23, 2026
Go to the source
Vercel Blogvercel.com
Publisher excerpt: DeepSeek V4 is now available on .Vercel AI Gateway There are 2 model variants: DeepSeek V4 Pro and DeepSeek V4 Flash. A 1M token context window is the default across both models. DeepSeek V4 Pro focuses on agentic coding, formal mathematical reasoning, and long-horizon workflows. It handles feature…