A Token Plan for AI models. One subscription, one OpenAI-compatible endpoint — and you never touch an upstream provider account again. Higher plans unlock the highest-scoring model band.
Each plan is a quality band, and they nest — a higher plan includes everything below it.
One base URL, one key. Point any OpenAI-compatible client at it and it just works.
Public endpoint info below — your key is issued in the console after you sign in.
Split by modality. Use the aggregate name — you never memorise upstream names.
Anything that accepts a custom OpenAI-compatible endpoint. No plugin needed.
Same setup everywhere: set the base URL to https://alli.website/v1, use your key, and pass any model name from the list above.
The free tier issues a key on signup — no card needed. Paid tiers activate immediately.
Change the endpoint in your tool to https://alli.website/v1 — nothing else changes.
Use fm-v1-* for chat, and auto/* for image / video / and the rest. Full catalogue below.
Loading…
Six provider accounts, six bills, six rate limits — or one.
Every model behind a single OpenAI-compatible base URL. Swap models by changing one string — no SDK, no provider onboarding, no per-vendor keys.
The paid aliases route across a pool ordered by model score. If an upstream is down or rate-limits you, the request rolls to the next model instead of failing.
The highest-scoring models sit in the top band. Lite is free forever; Standard and Pro unlock the bands above it.
Per-model and per-provider token counts in the console, so you know what your spend is actually buying.
Non-chat models are not part of any plan — they are free for every account while the current promotion runs.
A subscription to a quality band of models rather than to a provider. You get one API key that calls every model in your plan, plus everything in the plans below it. Plans nest: Pro includes all of Standard, which includes all of Lite.
No — that is the point. We hold the upstream accounts. You only ever use the key you get here.
Yes. Point any OpenAI SDK at our base URL and pass your FreeModel key. Streaming, tool calls and the usual parameters work as expected.
Lite is free forever and needs no card — 23 models, including several coding and long-context models. Subscribe only when you need the higher bands.
Two families. Chat: fm-v1-lite / fm-v1-standard / fm-v1-pro — one name per plan, routed across a pool ordered by model score, so you never hardcode an upstream name. Non-chat: auto/image, auto/video, auto/tts, auto/asr, auto/embed, auto/rerank. The full directory is above; response bodies always tell you which concrete model actually served.
Base URL https://alli.website/v1, auth header Authorization: Bearer <your key>. Anything that speaks the OpenAI API works — Cursor, Cline, Continue, the OpenAI SDK, LiteLLM, Cherry Studio, one-api. No plugin needed.
Yes. Upgrades take effect immediately and the unused time on your current plan is credited. Yearly simply bills ten months for twelve.
Handing off from FreeModel. This only takes a moment.
Your key is stored only in this browser (localStorage) and sent to this API to identify you. It is never sent anywhere else.