Changing a base URL is only one part of the setup. Someone also needs to own provider access, key permissions, routing decisions and the costs behind the connection.
Before you connect anything
Make an inventory before connecting.
01Tool and compatibility
Which application needs model access? Does it support the OpenAI-compatible endpoints you need?
02Provider and key owner
Who supplies provider access and manages the issued Gateway keys? Do not put keys in this inventory.
03Routing responsibility
Who decides permitted models, capabilities and residency policy?
04Budget and usage owner
Who reviews recorded usage, budget settings and provider charges?
Make the decision tangible
An inventory with explicit owners.
Illustrative setup inventory · fictional tools and roles. No credentials, live usage, savings or provider availability are represented.
Developer tool
A coding client.
Compatibility
Confirm its API base URL format and required chat-completion behavior.
Access owner
Platform administrator: issues the Gateway key and checks provider access.
Policy owner
Engineering lead: agrees permitted model routes for this tool.
Usage reviewer
Engineering operations: reviews attributed usage and provider charges.
Internal application
A document-summary job.
Compatibility
Check the application’s model client and the exact endpoints it needs.
Access owner
Application owner: manages the application credential securely.
Policy owner
Platform administrator: configures the relevant key restrictions.
Usage reviewer
Service owner: reviews usage, rate limits and the agreed budget settings.
A practical next step
Connect, check, then generate.
Agree the setup
Confirm tool compatibility, provider access, key ownership and policy. A website visit does not create access.
Check an issued connection
Use the existing first-connection guide with your supplied API base URL and Gateway key. The catalogue check does not generate an answer.
Inspect the ongoing responsibility
Agree software, model, hosting and support costs and who reviews usage before wider adoption.
Where Odin fits
Connect the parts this work needs.
Choose the capabilities around the task. These are configured starting points, not an automatic bundle.
Compatibility and provider availability depend on the setup. Residency restrictions apply to configured Gateway chat-completion policies, not automatically to every product or request. Odin Next model endpoints are configured separately from Gateway keys.
Move forward with a clear question
Already have connection details?
Use the catalogue guide only with an issued Gateway key and supplied API base URL. Without access, bring the inventory above to a setup conversation; no key needs to be shared here.
