Plans & credits
Your subscription gives you a monthly budget of credits. Every plan can use every model — nothing is locked behind a higher tier. What changes is how fast a model spends your credits.
Why credits instead of tokens
A credit reflects what a request actually costs to run. A lightweight model barely touches your budget; a frontier model spends it quickly. That means you keep the freedom to reach for the best model when a task needs it, without paying for that capability on the tasks that don't.
Roughly, on a typical chat message:
| Model | Credits per message |
|---|---|
| Gemini Flash | about 1 |
| GPT-4.1 mini | about 2 |
| GPT-4o | about 15 |
| Claude Sonnet 4 | about 20 |
The exact rate for every model is shown on the pricing page, and as a small cr/1k badge next
to each model in the model picker — so you can see the cost before you choose, not after.
Choosing a plan
| Free | Individual | Pro | Business | |
|---|---|---|---|---|
| Price | 0 | €19/month | €59/month | €39 per seat/month, from 3 seats |
| Credits per month | 500 | 5 000 | 20 000 | 15 000 per seat, shared |
| Projects | 1 | 5 | 25 | Unlimited |
| Agents | — | 3 | 15 | 100 |
| MCP servers | 1 | 3 | 10 | 25 |
| Integrations | — | 3 | 15 | Unlimited |
| Automations | — | — | 10 | Unlimited |
| Scheduled tasks | — | 3 | 25 | 100 |
| Knowledge bases | 1 | 3 | 15 | Unlimited |
| Document storage | 50 MB | 1 GB | 10 GB | 50 GB |
| API access & keys | — | — | 5 keys | 25 keys |
| Realtime voice | — | — | ✓ | ✓ |
| Your own provider key | — | — | ✓ | ✓ |
Annual billing is charged for ten months instead of twelve.
On Business, credits are shared across the whole organization. One person having a heavy week does not have to be balanced against everyone else's quiet one — the team draws on a single pool.
Watching your usage
Settings → Billing shows, for the current month:
- credits used against the credits included in your plan,
- how many purchased credits you still hold,
- how much total headroom is left,
- your usage of every other limit — agents, MCP servers, integrations, storage and so on.
Usage resets on the first day of each calendar month.
Running out
If you reach your monthly budget, requests that need a model stop with a clear message rather than failing silently. You have two options.
Buy a credit pack
From Individual upwards you can buy extra credits at any time:
| Pack | Credits | Price |
|---|---|---|
| Small | 2 500 | €10 |
| Medium | 10 000 | €39 |
| Large | 25 000 | €95 |
Purchased credits are used only once your monthly allowance runs out, and they stay valid for 12 months — so buying a pack near the end of a month is never wasted.
Turn on auto-recharge
Rather than being interrupted mid-task, enable auto-recharge in Settings → Billing: when your remaining credits fall below a threshold you choose, a pack is purchased automatically. It is off by default and only the organization owner can change it.
If you find yourself topping up every month, upgrading is better value — the plans include credits at a lower unit price than the packs do.
Getting more out of your budget
- Match the model to the task. Drafting, summarising and classifying rarely need a frontier model.
The
cr/1kbadge in the model picker makes the difference visible at the moment you choose. - Keep prompts and context tight. Input tokens are charged too, and long transcripts add up across a conversation.
- Watch scheduled tasks and automations. They spend credits on their own schedule, without anyone watching. Review them in Settings → Billing if a month runs hotter than you expected.
Who can change billing
Only the owner of the organization can subscribe, change plan, adjust seats, buy credits or configure auto-recharge. Other members can see usage but not change the subscription.
Need negotiated volume, SSO or an SLA? Contact us — those are handled as a custom plan rather than through self-serve checkout.