Summary

On July 28, 2026, Cloudflare began requiring the Workers Paid plan to run three resource-intensive models on Workers AI: @cf/moonshotai/kimi-k2.6, @cf/moonshotai/kimi-k2.7-code, and @cf/zai-org/glm-5.2. Free-plan requests to these models now return an HTTP 403 (internal error 5035). Cloudflare framed the change as protecting capacity for the broader Workers AI user base; lighter models such as GLM-4.7-Flash, Gemma 4 26B, and Nemotron 3 120B remain free.

What changed

Three models (kimi-k2.6, kimi-k2.7-code, glm-5.2) now require the Workers Paid plan, which starts at $5/mo and still includes 10,000 free daily Neurons. Free-tier calls to them return HTTP 403 / internal error 5035. Other capable models stay on the free tier.

Why it matters

Signals that serverless-inference economics are tightening: the most compute-heavy open models are moving behind a paid gate as demand grows. Developers prototyping agentic or coding workloads on the free tier with Kimi K2 or GLM-5.2 must now upgrade or switch models, and platform teams should expect capacity-based tiering to spread across inference providers.

Evidence excerpt

We are limiting Workers Free plan access to a few resource-intensive models so we can prioritize capacity for the broader Workers AI user base. The following models now require the Workers Paid plan: @cf/moonshotai/kimi-k2.6, @cf/moonshotai/kimi-k2.7-code, and @cf/zai-org/glm-5.2.

Sources