Infrintia API Reference

Base URL: https://api.infrintia.crossgl.net

Authenticated endpoints require a Bearer token or Basic Auth credentials in the Authorization header.

---

Public Endpoints

These endpoints require no authentication.

GET /

Platform information and version.

Response 200 OK

{
  "name": "Infrintia Compute Marketplace",
  "version": "1.0.0",
  "docs": "https://crossgl.github.io/crossgl-docs/pages/infrintia/overview/"
}

GET /health

Health check.

Response 200 OK

{
  "status": "healthy",
  "active_hosts": 24,
  "queued_jobs": 3
}

GET /models

List all available models on the marketplace.

Response 200 OK

{
  "models": [
    {
      "name": "meta-llama/Llama-3-8B-Instruct",
      "available_hosts": 8,
      "min_price_per_token": 0.0001,
      "max_tokens": 4096
    }
  ]
}

GET /hosts

List all registered hosts and their status.

Response 200 OK

{
  "hosts": [
    {
      "host_id": "host_xyz",
      "gpu": "NVIDIA A100 80GB",
      "status": "active",
      "models": ["meta-llama/Llama-3-8B-Instruct"],
      "price_per_token": 0.00012,
      "backend": "huggingface"
    }
  ]
}

GET /metrics

Platform-wide metrics.

Response 200 OK

{
  "total_jobs_completed": 152340,
  "active_hosts": 24,
  "total_users": 1830,
  "avg_job_latency_ms": 3200
}

---

User Endpoints

POST /users

Create a new user account.

Request Body:

{
  "username": "alice",
  "email": "alice@example.com",
  "password": "secure-password-123"
}

Response 201 Created

{
  "user_id": "usr_abc123",
  "username": "alice",
  "email": "alice@example.com",
  "credits": 100.0,
  "created_at": "2025-06-01T12:00:00Z"
}

GET /users/me

Get the authenticated user's profile and credit balance.

Authentication: Required

Response 200 OK

{
  "user_id": "usr_abc123",
  "username": "alice",
  "credits": 87.50,
  "total_jobs": 42,
  "created_at": "2025-06-01T12:00:00Z"
}

POST /run/model

Submit an inference job.

Authentication: Required

Request Body:

{
  "model": "meta-llama/Llama-3-8B-Instruct",
  "input_text": "Explain quantum computing.",
  "max_tokens": 256,
  "temperature": 0.7
}

Field	Type	Required	Description
model	string	Yes	Model identifier (from GET /models)
input_text	string	Yes	Input prompt
max_tokens	integer	No	Maximum tokens to generate (default: 128)
temperature	float	No	Sampling temperature (default: 0.7)

Response 202 Accepted

{
  "job_id": "job_def456",
  "status": "queued",
  "model": "meta-llama/Llama-3-8B-Instruct",
  "credits_reserved": 2.56,
  "created_at": "2025-06-01T12:05:00Z"
}

GET /jobs

List the authenticated user's jobs.

Authentication: Required

Query Parameters:

Parameter	Type	Description
status	string	Filter by status: queued, running, completed, failed
limit	integer	Results per page (default: 20)
offset	integer	Pagination offset (default: 0)

Response 200 OK

{
  "jobs": [
    {
      "job_id": "job_def456",
      "model": "meta-llama/Llama-3-8B-Instruct",
      "status": "completed",
      "credits_used": 1.92,
      "created_at": "2025-06-01T12:05:00Z"
    }
  ],
  "total": 42
}

GET /jobs/{id}/stream

Stream job results via Server-Sent Events (SSE).

Authentication: Required

Response: text/event-stream

data: {"token": "Quantum"}
data: {"token": " computing"}
data: {"token": " is"}
data: {"token": " a"}
data: [DONE]

Each event contains a single generated token. The stream ends with [DONE].

---

Host Endpoints

POST /hosts/register

Register a new GPU host on the marketplace.

Request Body:

{
  "gpu": "NVIDIA A100 80GB",
  "models": ["meta-llama/Llama-3-8B-Instruct"],
  "price_per_token": 0.00012,
  "backend": "huggingface",
  "endpoint_url": "https://my-gpu-host.example.com/infer"
}

Response 201 Created

{
  "host_id": "host_xyz",
  "status": "active",
  "api_key": "hk_live_xxxxxxxx"
}

POST /hosts/heartbeat

Send a heartbeat to confirm the host is still active. Should be called every 30 seconds.

Authentication: Required (Host API key)

Request Body:

{
  "host_id": "host_xyz",
  "gpu_utilization": 0.45,
  "available_models": ["meta-llama/Llama-3-8B-Instruct"]
}

Response 200 OK

{
  "status": "acknowledged"
}

POST /hosts/next-job

Poll for the next available job assigned to this host.

Authentication: Required (Host API key)

Request Body:

{
  "host_id": "host_xyz"
}

Response 200 OK (job available)

{
  "job_id": "job_def456",
  "model": "meta-llama/Llama-3-8B-Instruct",
  "input_text": "Explain quantum computing.",
  "max_tokens": 256,
  "temperature": 0.7
}

Response 204 No Content — No jobs available.

---

Admin Endpoints

These endpoints require admin-level authentication.

GET /admin/users

List all users on the platform.

Response 200 OK

{
  "users": [
    {
      "user_id": "usr_abc123",
      "username": "alice",
      "credits": 87.50,
      "total_jobs": 42,
      "created_at": "2025-06-01T12:00:00Z"
    }
  ],
  "total": 1830
}

GET /admin/hosts

List all registered hosts.

Response 200 OK

{
  "hosts": [
    {
      "host_id": "host_xyz",
      "gpu": "NVIDIA A100 80GB",
      "status": "active",
      "total_jobs_completed": 1240,
      "total_earnings": 148.80,
      "last_heartbeat": "2025-06-01T12:10:00Z"
    }
  ],
  "total": 24
}