Documentation Menu

API Reference

Developer trust is not built with sample cURL alone. Stable contracts, a clear error model, predictable versioning, and explicit protocol behavior form the real trust base of a good API. This page is designed so you can read the BriqMind API structure as a technical contract surface.

01Core Endpoints

A good API reference is more than an endpoint list; it is a behavior contract between the client and the platform. The main protocol routes you can use to access the system are defined below.

POST/v1/chat/completions

Main entry point for synchronous response generation, tool calling, and multi-step tasks.

POST/v1/responses

Extended surface for long workflows, more flexible orchestration, and job-based usage.

GET/v1/models

Base discovery endpoint for accessible models and deployments.

POST/v1/embeddings

Embedding generation surface for RAG, search, and similarity-based retrieval flows.

POST/v1/images/generations

Image generation, variation, and brief-based creative output with the Blink model.

POST/v1/audio/speech

Text-to-speech, voice responses, and speech-based output with the Blip model.

02API Contracts

Request and response structure, stream flows, and retry standards help developers build fault-tolerant applications.

Request Contract

Required fields, optional parameters, model selection, tool definitions, and validation rules should be explained as one contract.

Response Contract

Successful responses, the error model, metadata fields, and tool-output structure should be stable and predictable.

Streaming Protocol

Real-time event flow, token events, finalization, and client close behavior should be documented as a separate protocol.

Retry & Idempotency

How repeated requests are handled, duplicate jobs are prevented, and safe retry behavior works must be clear.

03Reference Layers

A good reference helps developers understand the protocols around endpoints as much as the endpoints themselves. Supporting layers run from security to observability.

Authentication

Bearer tokenEnvironment scopeService account usageKey rotation

Versioning

Breaking change policySunset announcementsMigration notesVersion path strategy

Error Model

Code / type standardRemediation suggestionRetryability informationExample response body

Observability

Request IDRate limit headersLatency measurementWebhook / async traceability

04Request Lifecycle

1. Incoming request

The client sends a secure request with model, messages, and optional tool definitions.

2. Contract validation

Auth, payload structure, model access, and parameter compatibility are evaluated on the server side.

3. Processing and orchestration

The request is routed to the right model or orchestration flow; tool calling and internal logic run here.

4. Response or event flow

The result returns to the client either as a synchronous response body or through a streaming / async event model.

Reference Note

A strong API reference lowers integration risk for developers. That does more than create technical satisfaction; it shows the platform's seriousness, maturity, and readiness to integrate with larger systems.