Cancel anytime. No surprises.
One full plan per seat. Each teammate gets their own 15M-credit pool.
Kick the tires. No card required.
Billed $119.88 annually · save 33%Billed monthly · cancel anytime
Pay once. 15M tokens / month. Forever.
Cheaper than ChatGPT alone — and you get every other AI too.
Need extra tokens this month? Just buy a top-up pack — no waiting, no emails.
One pool for every model. Cheap models burn fewer tokens, frontier models burn more — you decide where to spend.
| Model | Tokens / typical message | Messages per Pro month |
|---|---|---|
| Llama 3.3 70B · DeepSeek V3 · Mistral Small | ~235 – 590 | 25,000 – 64,000 |
| Gemini 2.0 Flash · GPT-4o mini · Claude Haiku | ~375 – 1,180 | 13,000 – 40,000 |
| GPT-4o · Claude 3.5 Sonnet · Grok 2 | ~3,750 – 9,200 | 1,600 – 4,000 |
| o1 · o3 · Claude Opus 4 · Gemini 2.5 Pro | ~22,500 – 90,000 | 165 – 650 |
Estimates only — actual cost depends on prompt + response length. Every assistant message shows the exact tokens charged, so there's no guessing.
No mystery tokens, no per-tool pricing, no “talk to sales”.
Top up any time. Tokens never expire while you have an active plan.
A token is one unit of inference. Cheap models like Llama 3.3 70B or DeepSeek V3 burn a few hundred tokens per message; frontier models like Claude Opus or o3 burn tens of thousands. The Pro plan's 15M-credit pool covers about 3,900 GPT-4o messages or 39,000 GPT-4o-mini — pick the right model per turn and the pool stretches further. Every assistant message shows the exact tokens charged.
Pro plan: your monthly pool refills on each renewal — unused tokens do not roll over. Top-up packs do stack and stay until you spend them (or your plan lapses).
The composer tells you before you send a message that's too expensive. Pick a cheaper model, buy a top-up, or wait for your renewal. We never silently fail or downgrade your model.
Open-weight models like Llama 3.3 70B, DeepSeek V3, and Mistral Small 3.1 are a small fraction of the cost of GPT-4o or Claude Opus — a few hundred tokens per message instead of several thousand. For most everyday work they're great and your pool lasts much longer.
Every assistant message shows the exact tokens charged underneath, and your remaining balance updates in the toolbar in real time.
Yes. It's drop-in compatible with the OpenAI SDK (just point base_url at https://api.faceb.ai/v1) and uses your same token balance. One wallet for both the chat and the API.
Open your account page and click “Cancel subscription”. You keep Pro access until the end of your billing period; nothing else is required.
Yes. Email hello@faceb.ai within 7 days of your first paid charge and we refund in full, no questions asked. Lifetime is covered too.
No, we don't train on your chats and we don't sell your data. Conversations are stored only so you can see your history and so we can charge the right tokens. The major upstream API providers (OpenAI, Anthropic, Google, Meta) also don't train on API traffic by default — see each provider's API terms if you're handling sensitive data.
Uploading is free. The tokens charged are whatever the model bills for processing the input — images add tokens to the prompt, PDFs and text files are extracted to plaintext on our side first. The cost shown under each assistant message includes everything the model saw.
Annual Pro is $119.88/year, so Lifetime ($499) breaks even at ~4 years. If you know you will use AI heavily for the long haul, Lifetime is the cheaper deal forever. If you are still figuring out your usage, start on annual Pro — you can switch to Lifetime any time later.
All major token and debit cards via Stripe (Visa, Mastercard, Amex, Discover, JCB, Diners), plus Apple Pay and Google Pay where supported. We don't store your card — Stripe does.