Every piece needed to build a full application on Cloudflare’s edge, read one at a time: the Workers execution model, where D1, KV and R2 each stop being the right answer, the framework layer of Hono and OpenNext, then Workers AI bindings and the domain and native-module problems that show up at deploy time.
The Workers AI catalog currently holds 84 models. For general chat pick glm-4.7-flash ($0.06 / $0.40 per M, 131K context), for vision pick gemma-4-26b-a4b-it ($0.10 / $0.30, 256K), for cheap high-volume steps pick granite-4.0-h-micro ($0.017 / $0.112), and for embeddings pick qwen3-embedding-0.6b or bge-m3 (both $0.012 per M). This post is updated on a schedule.
Cloudflare Custom Error Pages require a paid plan. On the Free Plan, use a Worker with inline HTML to intercept 5xx responses instead.
In wrangler.jsonc, use custom_domain: true in routes with only the hostname as the pattern — no /* wildcard
Even when a route has prerender = true, Cloudflare Workers' Rollup bundler still attempts to bundle native modules, causing the build to fail. The fix is to move any native module work into a postbuild script.