Providers

Aggregator with a free plan and a small starter token allocation.

OpenAI- and Anthropic-compatible gateway that subsidises a set of *-free model ids at its own cost.

Per-model token grant with a ninety-day window and no identity check needed.

Vendor-run free model API with both OpenAI and Anthropic dialects.

Auto-routing gateway that prefers $0 upstream capacity before paid fallbacks.

Community events that open one enormous shared token pool on a frontier model for every account.

Per-model token grants across seventeen models, time-boxed.

API access stays inactive until a verified payment method is added.

Confidential-compute inference. Free tier retired 15 March 2026.

Edge inference metered in neurons, with a daily free allocation.

Trial API access, explicitly restricted to non-commercial use.

A small signup credit, cardless, on prepaid billing.

Discord-keyed public API serving donated and sponsored capacity across several upstream providers.

Retired 30 July 2026. Retained as a historical record.

Gemini models free of charge on a cardless tier, with the longest context of any first-party offer.

LPU inference at very high token rates on a cardless free plan.

Inference router fronting many upstream providers against a monthly credit.

Community cluster of donated idle GPUs. The public pool costs no cash, only compute.

Keyless per-IP gateway route over a shared pool of free model capacity.

Key-optional donor-funded tier reaching a small set of open models.

Standing free mode with hidden limits readable only in the admin panel.

Daily call allowance across a community model catalogue.

Sign-up voucher valid three months, excluded from the newest model.

Free trial suspended 13 July 2026. Builder Program still grants expiring credit.

Two models priced at zero in and zero out.

Vendor-hosted open-model catalogue with a cardless prototyping rate limit.

The full frontier catalogue with a subset of ids served free inside the coding agent.

Model router whose :free variant costs zero per input and output token.

Community endpoint with an anonymous keyless text route and a Pollen-metered catalogue.

RDU inference on a cardless tier defined as having no payment method linked.

Sign-up voucher plus a set of genuinely zero-cost embedding and speech models.

Public pool of frontier-model tokens funded by third-party sponsors.

Free trial ended 18 July 2026. Paid only.

A million shared tokens for a year across all language models.

Free tier removed July 2025; a minimum credit purchase is now required.

Free model id on a separate daily allowance from the paid key limit.

Monthly gateway credit that requires a payment method before it can be spent.

Per-model token grants, plus a programme that pays credit for inference data.

GLM Flash models priced free at every token position, input and output alike.