FAQ

What is the Accly Gateway API?

Accly Gateway is a multi-model AI gateway API for accessing supported AI models through one Accly account.

Use it from your server, agent, automation, CLI tool, or backend application with an Accly API key. The gateway supports familiar API formats, including OpenAI-compatible chat completions, Anthropic-style messages, and Google/Gemini-style endpoints where available.

Does Accly Gateway support other API formats?

Yes. The primary quickstart endpoint is OpenAI-compatible chat completions, and Accly Gateway also supports Anthropic-style and Google/Gemini-style endpoint families where available.

See Gateway API for request and response examples.

Which API key type should I choose?

For most OpenAI-compatible tools, choose openai or universal. Choose anthropic if your integration uses Anthropic-style Messages API requests. Choose google if your integration uses Google/Gemini-style generate content requests.

Universal is for OpenAI-compatible multi-model routing; it does not guarantee every provider-native feature.

Do I need separate OpenAI, Anthropic, or Google API keys?

No. For normal Accly Gateway API usage, you use an Accly API key. Choose the key type that matches your endpoint family.

Which models can I use through the API?

Supported models are the active models shown in the Accly model picker or account model list.

Model availability can change. Plans control access by tier: Basic, Advanced, Thinking, and Beta.

Can I use Accly with coding tools and agents?

Yes. Use the Gateway API from your server, CLI tool, agent, or automation when your plan and selected model support the workflow.

Tools that support custom OpenAI-compatible base URLs usually use openai or universal keys. Provider-native clients may require the matching key type.

Do not put Accly API keys in public frontend code.

Can I use Accly commercially?

Yes, within your plan limits and Accly’s Terms, AI Usage Terms, and Acceptable Use Policy. You are responsible for how your application uses model output.

Can I build a public AI service on top of Accly?

Accly plans are intended for individual/team usage and automation workflows, not resale-style high-throughput infrastructure unless explicitly permitted by Accly.

If you build a public-facing product, you are responsible for your user controls, abuse prevention, generated output handling, and compliance with Accly’s policies.

What limits apply to API usage?

Gateway API usage can be limited by:

• API key access
• Model tiers
• Daily request units
• Request-count guards
• RPM rate limits
• Subscription or account status

See Pricing & Usage.

What are request units?

Request units are the daily usage allowance for paid plans. One API call does not always equal one unit.

Lightweight model requests can consume one unit. Heavier models, larger context windows, long outputs, cached context reads, and agentic coding workflows can consume more units.

Do request units have cash value?

No. Request units are usage capacity for Accly plans. They are not transferable and cannot be redeemed for money.

Does Accly replace the model I choose?

No. Accly does not silently replace your selected model with an unrelated cheaper model.

If a model route is degraded, delayed, unavailable, or economically unsafe to serve, Accly can limit, queue, or temporarily pause that route instead of changing the model behind your back.

Why does Max have the same RPM as Pro?

Max is designed for higher daily capacity, not higher resale-style throughput.

It includes more request units and a larger request-count guard while keeping RPM controlled.

What happens when I hit my limits?

The gateway returns an error with a specific error.code.

Common examples:

• rate_limit_exceeded
• request_units_exceeded
• daily_limit_exceeded
• free_balance_exhausted
• model_not_allowed

Some legacy clients or cached errors may still show budget_exceeded during the request-unit migration.

See Errors & Limits.

Are prompts and outputs sent to third-party AI providers?

Yes. Prompts, uploaded content, and model outputs can be sent to third-party AI providers to generate responses.

Model availability and capabilities may change.

Are chats or API requests used for training?

Accly does not use customer chats or Gateway API requests to train Accly-owned foundation models.

Prompts and outputs may still be processed by third-party AI providers to generate responses.

Can I delete or export my data?

The app includes account data controls, including export and deletion request flows.

Some billing, security, fraud-prevention, legal, or abuse-prevention records may need to be retained where required or reasonably necessary.

How do billing and refunds work?

Refunds are reviewed case by case through Accly support and the active checkout provider.

Refund policy:

https://accly.net/en/refunds