Free AI Models

Yes. All 48 models bill $0 for input and output tokens. AIHubMix subsidizes the inference cost; usage is bounded by quotas instead of a bill.

10 trial calls, shared across all free models, with no expiry. When they run out the API keeps answering with a note that the trial is used up — your integration never sees a hard error.

Any top-up of $1 or more permanently switches your account to daily quotas: 10 requests per minute, 100 requests per day and 1M tokens per day, shared across the free catalog and reset every day. Models with a higher request weight count as several requests per call — the exact caps are in the table. Past a cap the API returns 429 until the next minute or day starts.

Each call to a higher-weight model counts as several requests against the daily 100. The Rate limit column shows the resulting effective caps per model.

They are the same upstream model. The -free ID runs as the subsidized free variant with the limits above; the paid ID bills per token with no daily cap. Switching between them is a one-line model-name change.

Yes. Free models sit behind the standard OpenAI-compatible endpoint, so the official SDKs, LangChain, LlamaIndex and any client with a custom base URL work unchanged.

Moderate production traffic is fine. The pattern that works: paid models on the critical path, free models for auxiliary work like batch summarization, drafts, and log analysis.