JA EN
shiichan

Some Workers AI Models Just Went Paid-Only!

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

Hey, it's me, Shiichan! Today I found an update to Cloudflare Workers AI's free plan. Depending on which model you use, this could affect you, so let's take a look.

Cloudflare Changelog developers.cloudflare.com

What was announced?

The Cloudflare Changelog announced that some Workers AI models are no longer available on the Workers Free plan and now require the Workers Paid plan. The affected models are:

  • @cf/moonshotai/kimi-k2.6
  • @cf/moonshotai/kimi-k2.7-code
  • @cf/zai-org/glm-5.2

Starting July 28, 2026, requests to these models on the Workers Free plan return a 403 error (internal error 5035) prompting you to upgrade.

The story so far

Until now, the Workers Free plan gave broad access to Workers AI models, including resource-intensive ones. According to Cloudflare, that put pressure on capacity for popular models, leading to more 429 errors and 3040 errors (Out of Capacity).

What changes

  • Free plan users now need to upgrade to Workers Paid to keep using the three affected models
  • Many other models (like @cf/zai-org/glm-4.7-flash, @cf/google/gemma-4-26b-a4b-it, and @cf/nvidia/nemotron-3-120b-a12b) remain available on the Free plan
  • By limiting access to the heaviest models, Cloudflare expects fewer 429 and 3040 errors across Workers AI overall, for a more reliable inference experience

So if you were using the listed models on the Free plan, this affects you directly. For everyone else, it's also an update that should mean less congestion and smoother access.

Dive Deep

The Workers Paid plan starts at $5 per month and still includes the 10,000 free Neurons per day allocation, with usage beyond that billed at each model's pricing. You can check the full model list in the Workers AI model catalog.

Wrap-up

  • kimi-k2.6, kimi-k2.7-code, and glm-5.2 now require the Workers Paid plan
  • Free plan requests return a 403 error (internal error 5035) prompting an upgrade
  • The Paid plan starts at $5/month with 10,000 free Neurons per day
  • The goal is to ease capacity pressure and cut down on 429 and 3040 errors overall
  • Worth checking if you were relying on these models for free; otherwise, this should mean a more stable experience for everyone.

Related posts