Summary
Cloudflare now requires the Workers Paid plan (from $5/month) to run three resource-intensive Workers AI models: Kimi K2.6, Kimi K2.7-Code, and GLM-5.2. Cloudflare frames the change as capacity management to cut 429 and out-of-capacity errors; Workers Free users calling these models get a 403 with an upgrade prompt, while many smaller models such as GLM-4.7-Flash and Gemma-4-26B stay free.
What changed
On July 28, 2026 Cloudflare restricted Kimi K2.6, Kimi K2.7-Code, and GLM-5.2 on Workers AI to the Workers Paid plan, returning 403 errors with upgrade prompts to Free-tier callers, citing more reliable inference capacity.
Why it matters
Gating the heaviest open-weight models behind a paid tier aligns cost and capacity as inference gets more expensive to serve; it signals Cloudflare tiering Workers AI by model weight and steering serious agentic and coding workloads onto paid capacity while keeping lighter models free.
Evidence excerpt
Three resource-intensive models now require the Workers Paid plan: Kimi K2.6, Kimi K2.7-Code, and GLM-5.2. Workers Free users receive 403 errors prompting upgrades.