JA EN
shiichan

DeepSeek V4 lands on Workers AI with a full 1 million token context window!

Cloudflare Changelog ★★★★★ View Markdown
#cloudflare#ai#serverless

Hey everyone, it's me! There's a big new arrival on Workers AI today, and it's bringing a seriously long context window with it.

Cloudflare Changelog developers.cloudflare.com

What was announced?

On the Cloudflare Changelog, Cloudflare announced that DeepSeek V4 Pro and DeepSeek V4 Flash are now available on Workers AI. Their model IDs are @cf/deepseek-ai/deepseek-v4-pro-0813 and @cf/deepseek-ai/deepseek-v4-flash-0731. These are the first models on Workers AI with a full one million (1,048,576) token context window.

Why it matters

No model hosted on Workers AI has ever had a context window this large before. Long-horizon agentic workflows, loading entire large codebases, and multi-step reasoning that would normally hit other models' context limits can now be handled in full by these two models.

What changes

With DeepSeek V4 Pro and Flash live on Workers AI, these key capabilities are now available:

  • Reasoning: both models support thinking mode for complex, step-by-step problem-solving
  • Function calling: build agents that invoke tools and APIs across multiple conversation turns
  • Long context: both models support the full one million (1,048,576) token context window

There are several ways to access them too:

  • The Workers AI binding (env.AI.run())
  • The REST API
  • The OpenAI-compatible endpoint
  • AI Gateway

Dive Deep

DeepSeek V4 Flash is positioned as the faster, lower-cost sibling of the Pro model, so you can pick whichever fits your use case.

This release supersedes the earlier preview version, with substantially enhanced agentic capabilities.

To use either model, you'll need the Workers Paid plan or prepaid AI Gateway credits — they aren't available on the free plan yet.

For full specs and pricing, check the DeepSeek V4 Pro model page, the DeepSeek V4 Flash model page, and the pricing page.

Wrap-up

  • DeepSeek V4 Pro / Flash are now live on Workers AI
  • Both are Workers AI's first models with a full 1 million token context window
  • They bring reasoning, function calling, and long context as their three core capabilities
  • You'll need the Workers Paid plan or AI Gateway credits to use them
  • Access them via the Workers AI binding, REST API, OpenAI-compatible endpoint, or AI Gateway

This one's a great fit for developers building agents over large codebases, or anyone who wants long, multi-step reasoning running right at the edge!

Related posts