Skip to content

quickstart.md

Latest commit

 

History

History
102 lines (72 loc) · 3.85 KB

quickstart.md

File metadata and controls

102 lines (72 loc) · 3.85 KB

Consumer Quickstart

Send your first request to Darkbloom's OpenAI-compatible API.

Prerequisites

  • A Darkbloom account. Sign up at console.darkbloom.dev.
  • An API key. Create one in Settings → API Keys in the console, or via POST /v1/keys authenticated with a Privy JWT (see authentication.md).

Base URL

https://api.darkbloom.dev/v1

Authentication

Send a Bearer token in the Authorization header. requireAuth accepts either an API key (sk-db-...) or a Privy JWT (eyJ...), but the console/web flows require Privy and sensitive key-management endpoints reject API keys entirely (coordinator/api/server.go:1776-1888).

curl -H "Authorization: Bearer YOUR_API_KEY" \
     https://api.darkbloom.dev/v1/models

Your First Request

Python (OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.darkbloom.dev/v1",
    api_key="YOUR_API_KEY",
)

response = client.chat.completions.create(
    model="gemma-4-26b",  # use the ID returned by GET /v1/models
    messages=[{"role": "user", "content": "Hello, Darkbloom!"}],
    stream=True,
)

for chunk in response:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

cURL

curl -X POST https://api.darkbloom.dev/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemma-4-26b",
    "messages": [{"role": "user", "content": "Hello!"}],
    "stream": true
  }'

Anthropic Messages API

import anthropic

client = anthropic.Anthropic(
    base_url="https://api.darkbloom.dev/v1",
    api_key="YOUR_API_KEY",
)

response = client.messages.create(
    model="gemma-4-26b",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}],
)

Available Models

Models are loaded from the coordinator's DB-backed model registry, not hardcoded in the consumer docs. The authoritative list is always:

GET /v1/models

You can also use the provider-facing catalog at GET /v1/models/catalog. Aliases such as gemma-4-26b may resolve to different concrete builds over time; the consumer endpoint hides build details and exposes only the public alias (coordinator/api/model_alias_handlers_test.go:956-962).

Pricing

Darkbloom uses per-token pricing. Platform-set model prices are returned by GET /v1/pricing; if no platform price is configured, the fallback is $0.05 per 1M input tokens and $0.20 per 1M output tokens (coordinator/payments/pricing.go:24-29). Every charged request has a $0.0001 (100 micro-USD) minimum (coordinator/payments/pricing.go:31-32).

During public alpha the platform fee is 0%, so providers keep 100% of the per-token revenue (coordinator/payments/pricing.go:39-43).

Requests routed to your own provider machine via self-route are not charged. See self-route.

Next Steps