Frequently asked questions

Infer is a workspace for production AI agents. It combines OpenAI-compatible model access, usage visibility, and team billing behind one product surface.

Create a free account, generate an API key, add credits, and use the Infer base URL from the docs with any OpenAI-compatible SDK or HTTP client.

Yes. Infer is designed to work with OpenAI-compatible SDKs and HTTP clients by updating the base URL and API key.

Yes. Use `POST /v1/images/generation_tasks` to create an async image task and `GET /v1/images/generation_tasks/{task_id}` to poll it. The task envelope is Infer-specific, but it includes OpenAI-style `id`, `created`, and `status` fields, with `status` progressing through `queued`, `in_progress`, `completed`, or `failed`. Successful task results return OpenAI-style `data: [{ url }]` items.

Infer uses credit-based billing. Credits are consumed based on actual token usage, and each model has its own per-token price in the Models catalog.

AI model pricing may adjust over time because of upstream provider pricing, compute demand, model optimization, and infrastructure costs.

Today we accept USD1 stablecoin deposits. Card payments and additional payment options are on the roadmap.

Async image tasks run the final image generation through the same LiteLLM image runtime as synchronous image requests. The team budget is checked before queueing, and the completed generation is attributed to the creating team, user, and API key. The task ID is used as the billing idempotency key so crash recovery does not double-bill.

Models hosted in different geographic regions can experience cross-region latency, upstream outages, provider rate limits, or regional network filtering.

We monitor supported models and surface clear errors when a provider is degraded. Where supported, requests can be routed through healthier model/provider paths.

Yes. We support Server-Sent Events (SSE) streaming for compatible chat and completion models.

No. We do not store prompt or completion content. Usage metadata such as token counts, model IDs, and timestamps is logged for billing, rate limiting, and analytics.

The full privacy policy is available on the Privacy Policy page.

No. Once deposited or earned, credits remain in your account until they are consumed by API usage.

Credit purchases are final because on-chain transactions cannot be reversed. If a technical issue prevented use of purchased credits, contact support@infer.ai for a case-by-case review.

Check upstream provider status, try a different compatible model if the issue persists, and contact support for investigation when the problem continues.

For technical issues, account inquiries, or custom enterprise limits, email support@infer.ai.