Quickstart Guide
From zero to your first API call in under 5 minutes. Covers signup, key creation, and a working curl example.
Guide
Read guide →
OpenAI SDK Integration
Use the official OpenAI Python SDK with ModelHub by setting the base URL. Drop-in replacement for OpenAI calls.
Tutorial
OpenAI API docs →
Anthropic SDK Integration
Use the Anthropic SDK with ModelHub's /v1/messages endpoint. Same shape, same streaming, different provider.
Tutorial
Anthropic API docs →
Streaming Walkthrough
Everything you need to know about SSE streaming — chunk format, handling [DONE], and building a streaming UI.
Guide
Streaming docs →
API Key Best Practices
How to name, rotate, and secure your keys. Prefix-based identification, per-environment keys, and revocation strategy.
Best practice
API key docs →
Model Catalog & Pricing
Full reference for every model alias, its upstream mapping, and per-token pricing. Use GET /v1/models to list programmatically.
Reference
Models docs →
Trial & Billing Overview
How the $149 free credit works, how costs are calculated per request, and what happens when your balance runs out.
Guide
Billing docs →
Frequently asked questions
How do I get started?
Sign in with GitHub at /auth/login. Your first login creates an account with $149.00 free credit. Then go to /account to create an API key and start calling the API.
Which API shape should I use?
Use the OpenAI-compatible /v1/chat/completions if you have an OpenAI SDK or want the most widely supported format. Use /v1/messages if you're using the Anthropic SDK. Both work with the same API keys and billing.
What happens when my $149 credit runs out?
Requests return 402 Payment Required. An admin can grant more credit from the admin panel, or you can set up a payment integration. Your account remains active — you just can't make requests until the balance is positive.
How long does the trial last?
30 days from account creation by default (configurable via TRIAL_DAYS env var). After expiry, logged-out access to protected pages is blocked and you'll see the expired trial page. The trial is separate from your credit balance — you can have credit left but an expired trial.
Can I use multiple API keys?
Yes — create as many as you need from /account. Each key has a unique prefix (sk-project-random) for identification. Use separate keys for different environments (production, staging, local dev) so you can revoke one without affecting others.
Is streaming supported?
Yes — both /v1/chat/completions and /v1/messages support SSE streaming with stream=true. The response is text/event-stream. Each chunk is a JSON SSE frame, and the stream ends with data: [DONE].
What models are available?
Nine model aliases across three providers: OpenAI (gpt-6.1-terra $0.20/$0.60, gpt-6.1-sol $0.30/$1.10, gpt-6-astra $1.10/$5.10, gpt-6.1-luna $0.11/$0.15 — all 1.05M context), Claude (opus 5.5 $0.50/$2.10, sonnet 5.5 $0.30/$1.10, haiku 4.5 $0.20/$0.60 — 1M/200K context), and GLM/Zhipu (glm-5.3-prime $0.38/$0.98, glm-5.3-flashx $0.14/$0.23 — 1M context). All prices per 1M tokens. List them programmatically with GET /v1/models.
How is billing calculated?
Cost is estimated from the actual input and output token counts multiplied by the model's per-million-token price, divided by 1,000,000. The result is rounded down to the nearest cent and deducted from your balance.