Question 1

What is Gatewayz?

Accepted Answer

Gatewayz is the inference layer for AI agents: one API key and one model-agnostic endpoint for models from OpenAI, Anthropic, xAI, Moonshot and Meta. It is OpenAI-compatible at /v1/chat/completions and serves native Anthropic Messages at /v1/messages.

Question 2

Which AI models does Gatewayz support?

Accepted Answer

Models from OpenAI, Anthropic, xAI, Moonshot and Meta. The live list, with per-model pricing and context length, is always at https://api.gatewayz.ai/v1/models, so you never have to trust a number on a marketing page.

Question 3

What does "model agnostic" mean?

Accepted Answer

We don't sell a model of our own, so we have no reason to steer you toward one. You get the exact model you ask for: an undated alias resolves to its exact snapshot, and an unknown model id returns an error. We never quietly route your request to a different model and bill you for it.

Question 4

Why is Gatewayz built for AI agents?

Accepted Answer

An agent can't read an error message; it reads a status code. So every failure tells software what to do: an unknown model returns 400 model_not_found, a spent key cap returns 402 request_cap_exhausted, and an empty balance returns 402 insufficient_credits. None of them is a retryable 429 or 5xx, and a stream that fails upstream ends with an explicit error event instead of silently stopping.

Question 5

How does Gatewayz compare to OpenRouter?

Accepted Answer

Both give you one key across many models. Gatewayz is designed for unattended agents: native Anthropic Messages alongside the OpenAI-compatible API, machine-readable errors, per-key request caps, and a rule that we never substitute a model you didn't ask for.

Question 6

Is Gatewayz compatible with the OpenAI SDK?

Accepted Answer

Yes. Set the base URL to https://api.gatewayz.ai/v1 and use your Gatewayz API key. Switching models is a one-word change to the model id.

Question 7

Can I use Gatewayz with Claude Code?

Accepted Answer

Yes. Set ANTHROPIC_BASE_URL=https://api.gatewayz.ai and ANTHROPIC_AUTH_TOKEN to your Gatewayz key. Tool use and prompt caching pass through, so cache reads show up in the usage block as they would direct.

Question 8

How does pricing work?

Accepted Answer

Usage is billed per token from one credit balance across every provider, at the provider's list price plus a routing fee. The model catalog shows each provider's list price, so you can see what the underlying model costs rather than a blanket discount claim. You can cap each API key so an agent can't run past its budget.

Question 9

Do you store my prompts?

Accepted Answer

Not on plain API calls: we keep billing metadata (model, token counts, cost), not the content of prompts or completions. Chat history is stored only if you opt in to sessions.

Question 10

How do I get started?

Accepted Answer

Sign up on our website to get an API key and free credits, then point any OpenAI-compatible SDK or coding agent at https://api.gatewayz.ai.

openai/gpt-5.4

Pricing

Usage