Skip to main content
Use this guide when your application already sends OpenAI-compatible requests through Cloudflare AI Gateway. Move the gateway boundary first, then recreate only the routing, caching, and observability behavior your production traffic actually needs.

What changes

1

Inventory the current gateway configuration

Record the Cloudflare gateway URL, provider routes, custom headers, cache rules, retry behavior, and analytics your application relies on. This separates required behavior from settings that no longer need to move.
2

Replace the endpoint and credentials

Keep the request payload unchanged where possible and update the OpenAI-compatible client boundary.
3

Recreate routing and shared defaults

Move stable provider restrictions, prompt defaults, and generation parameters into Presets. Configure provider ranking and fallback behavior through Routing and Fallbacks.Cloudflare cache and analytics settings do not transfer automatically. Decide explicitly whether each behavior belongs in Phaseo, your application, or your edge layer.
4

Validate before cutover

Test non-streaming, streaming, errors, tools, and structured outputs as applicable. Compare latency, output, token usage, and cost before moving traffic gradually.

Migration checklist

  • Cloudflare-specific gateway URLs and headers have been removed.
  • Production model IDs are verified against GET /v1/models.
  • Required cache behavior has been deliberately recreated or retired.
  • Activity and request details provide the visibility your team needs.
  • Rollback remains a configuration-only endpoint and key change.

Next steps

Last modified on July 26, 2026