> ## Documentation Index
> Fetch the complete documentation index at: https://omnirelay.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Service boundaries

> Capabilities, activation and limits of the current live model service.

OmniRelay provides real text inference enabled per workspace. Query `GET /v1/models` with your own key for IDs, access, capabilities, output limits and customer prices. Catalog visibility, registration and top-ups do not activate models.

## Supported today

* Reviewed text models with JSON responses, plus SSE, system messages and reasoning content where enabled per model.
* GPT 5.4 Mini function tools, JSON object/schema and stateless Responses HTTP/SSE.
* Customer authentication, key creation/rotation/revocation, model allowlists and budgets.
* Usage records, diagnostics, CSV export, maximum holds and final settlement.

Most integrated text models allow up to 128,000 output tokens and default to 32,000 (Gemini 3.1 Pro Preview and Gemini 3.6 Flash: at most 65,536); DeepSeek V4 Pro, GLM 5.3 and Kimi K3 default to and allow at most 4,096; GPT 5.4 Mini defaults to 64 and allows at most 256. Read each model’s `capabilities` and `limits`; a brand does not establish feature support.

## Not enabled

* Strict function tools, vision, audio and image/video generation.
* No embedding model is enabled for the current provider group; the endpoint requires a reviewed input-only model contract.
* Sampling options, stored Responses, hosted tools and WebSocket.
* Idempotent chat retries, automatic paid retries or failover after ambiguous outcomes.
* Customer-supplied model provider credentials or automatic activation of new catalog models.

## Charges and failures

Real calls reserve a maximum charge, then settle validated final usage at frozen prices. Ambiguous dispatched outcomes can retain holds pending reconciliation; failure does not mean free. Top-ups do not raise key/member caps. See [Usage and billing](https://omnirelay.ai/docs/general/usage-billing.md), [Errors](https://omnirelay.ai/docs/general/errors.md) and [User manual](https://omnirelay.ai/docs/user-manual.md).

## Accounts and console

Supported features include browsing platform models before activation, verified-signup setup queues, existing-account activation requests, multiple saved sign-ins and login history. Account setup and funded inference readiness are checked separately. Cash and trial credits are tracked separately; calls use trial credits first, and cumulative top-ups remain separate from model consumption. See the [User manual](https://omnirelay.ai/docs/user-manual.md) for page help, mobile navigation and drawers. Operator-only actions are handled by OmniRelay support.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.