Tokens are the fuel your AI runs on.
Every message you send to an AI — and every reply it generates — is made up of tokens. A token is roughly a word, or part of a word. "Hello" is one token. "extraordinary" might be two. Your average message is about 50–200 tokens. The AI's reply is another 100–500.
Every time your assistant thinks, it burns tokens. The cost depends on which model is doing the thinking.
The cheaper the model, the faster and cheaper the token. The more capable the model, the more it costs per token — but the better the output for complex tasks.
This is the core problem with AI subscriptions today: you pay one flat rate for one model, and that model handles everything — whether the task needs it or not. Simple questions get expensive models. Complex work gets the same expensive model. Nothing is optimized.
Three Bears fixes this.
The right model for the right task. Automatically.
Three Bears is CE's model routing system. Every message you send is routed to one of three tiers based on what the task actually requires:
Efficient — fast, lightweight, and cheap. For quick questions, simple lookups, short summaries, and routine tasks that don't need deep reasoning. This is the model that handles most of what you do in a day without breaking the bank.
Balanced — the middle tier. For most substantive work: drafting, analysis, research synthesis, anything that needs more than a quick answer but doesn't require the full weight of a top-tier model.
Capable — the full capability tier. For complex reasoning, multi-step problems, nuanced writing, anything where the quality of thinking genuinely matters. This is the tier you want when the task is hard.
CE routes each message automatically. You can also switch tiers mid-conversation if you want more or less firepower on a specific message. Every message shows you the live token cost so you always know what you're spending.
Axiom — the system that picks the best model for each tier — runs daily. It evaluates the available models across the AI landscape, scores them on quality and cost, and selects the top performer for each tier. You always get the best model at each level, updated continuously.
One subscription. The whole landscape.
To access the best models from every major AI provider separately, you're looking at $60 a month — $20 each, three subscriptions, three apps, three chat windows, none of them talking to each other.
Contained Evolution routes across the world's best models — Gemini, Anthropic, OpenAI, and the full OpenRouter catalog — all through a single subscription. Three Bears picks the right model for each task automatically. You see the cost live. You switch tiers mid-conversation.
You don't need three subscriptions. You need one.
Efficient · Balanced · Capable — Three Bears picks the right model for every task automatically.
See exactly what you've spent. In real time.
Every message shows you the live token cost as it happens. Not a hidden counter — a visible number next to every reply. Efficient messages cost fractions of a cent. Balanced messages cost a little more. Capable messages cost more still, and they're worth it for the tasks that need them.
PASS users get a usage gauge — a gas gauge for your monthly token budget. You can see at a glance how much of your monthly budget you've used, broken down by tier. Check it anytime from the usage dashboard.
When you're running low, the app tells you before you hit the ceiling — not after. You can top up your usage budget directly from the app if you want more.
Free users see a daily message counter: how many you've used today, how many are left. It resets at midnight. No surprises.
What you get on each tier.
Free — always free
No credit card. No expiry. No bait and switch.
- Your personal assistant (Efficient model)
- Both Personal and Business sections
- All core apps: Notes, Scribble, Map It, Braindumps, Scope, Thought Bubbles, Boards, Files, Timeclock, Playbook, Bulletin Board, Team Chat, Daily Brief, The Village, Memory Island
- Memory Bank — 15MB storage, synced across devices
- Daily message counter — 150 messages/day, resets at midnight
- Hard cap: when you hit the limit, the app tells you clearly and resets tomorrow
The one ceiling: model choice. Free tier uses the Efficient model for all conversations. It's a real daily driver — not a demo.
PASS — $5/month
Everything in Free, plus:
- All three model tiers — Efficient, Balanced, Capable — switch mid-conversation
- Live token cost per message
- Axiom routing — best model per tier, updated daily
- 1GB Memory Bank storage
- Memory Bank MCP integration — connect to Claude.ai and other external AI tools
- Automations and scheduled tasks that run while you're away
- Monthly usage gauge — see your budget in real time
- Usage top-ups available when you need more
- Bring your own API key for additional usage from any provider (BYOK)
One subscription. One price. No per-seat surprises on personal use.
Already have an API key? Use it.
PASS includes CE's managed token budget — but if you have your own API key from Anthropic, OpenAI, or another provider, you can connect it to CE directly. Your requests route through your own key, against your own account with that provider. CE provides the platform and routing; you provide the key.
This is the option for power users who want unlimited usage without a per-message cost from CE. Connect your key in Settings → Integrations.
Start free. Upgrade when you're ready.
No credit card for the free tier. No commitment. Your assistant starts learning you from message one.