Question 1

What is Gatewayz?

Accepted Answer

Gatewayz is the inference layer for AI agents: one API key and one neutral 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 "neutral" 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.

Frequently asked questions

Models and routing

Billing

Data and security

Models and routing

Which providers can I reach?

Will Gatewayz ever send my request to a different model?

Which API surfaces are supported?

Billing

How am I charged?

Can I stop an agent from overspending?

What happens when my balance runs out?

Data and security

Do you store my prompts or completions?

How long do you keep request records?

Do you have an uptime SLA?