Atlas Inference · beta
Sign in to start a chat
Chat with the models your organisation can use on Atlas. Your organisation needs Atlas Inference enabled.
Use an access token
What would you like to work on?
Messages
Recent usage
Usage records contain model, status and token counts. They do not contain message text.
Connect to view usage.
Serving activity
Recorded request activity across all organisations. This is not a current model or service readiness check.
Operator administration
Manage product access using an immutable Atlas organisation ID. Creating a tenant does not create an Atlas account or grant Identity roles. Disabling access stops new requests; accepted requests can finish.
Organisations
Select an organisation to review its access and usage.
Global model availability
Availability applies to every organisation. Model endpoints and credentials remain server configuration.
Organisation usage
Aggregated across the organisation, without messages or individual user identities. Dates use UTC request start time. These are quota accounting records, not an invoice.
Connect with a token
Use an Atlas access token with permission to use inference. This page keeps it in this tab's memory until you disconnect or reload.
Use the API
Send requests to this origin. API clients include an Atlas access token in the Authorization: Bearer header. The local development API uses the test account without a token.
GET /v1/models- List the models your organisation can use.
POST /v1/chat/completions- Generate a response from the messages you send, without saving them. Set
stream: truefor server-sent events. POST /v1/conversations- Create an empty saved conversation.
POST /v1/conversations/{id}/messages- Generate a response with the conversation's saved context.
GET /v1/usage- Read your usage metadata.
The page, API and memory service deploy separately. This page sends requests to the configured API; it holds no model or storage credentials.