Kilo Code pricing for teams: what you pay for and how to control spend
A free platform for individuals, $15 per user on Teams, and inference at provider rates. The work is in watching the inference.
Kilo Code pricing has three parts: a platform plan that is free for individuals and $15 per user a month on Teams, AI inference billed at the model provider's rate, and cloud compute charged on top. For a team lead, inference is the part that moves month to month, so the real question is how you watch it; Kilo's own dashboard covers Kilo traffic, and UsageAtlas, a free open-source desktop app, adds a cross-tool view for developers who run several agents.
The three lines on the bill
Kilo's pricing page splits cost so you only pay for what you use:
| Part | Individual | Teams | Enterprise |
|---|---|---|---|
| Platform plan | $0 | $15 per user a month | Custom |
| AI inference | Provider rates | Provider rates | Provider rates |
| Kilo Pass (optional) | From $19 a month | From $19 a month | From $19 a month |
| Cloud compute | Pay as you go | Pay as you go | Pay as you go |
The extensions and CLI are open source under the MIT license, so nobody needs a seat just to try the agent. Teams adds usage analytics, centralized billing, shared BYOK and team management. Enterprise adds SSO, audit logs, model and provider limits, and SLAs.
Inference: four ways to pay
This is the variable cost, and Kilo gives you a choice:
- Free and local. Free models through Kilo Gateway, or local models through Ollama or LM Studio.
- BYOK. Use your existing Anthropic, OpenAI, Google or other provider keys and pay them directly. Kilo charges nothing for those requests.
- Kilo Gateway. Pay-as-you-go credits at exact provider rates, topped up when needed.
- Kilo Pass. A monthly credit subscription (Starter $19, Pro $49, Expert $199) that adds bonus credits on top.
A payment-processing fee applies when you buy credits or a Pass. Check the pricing page for the current figure before you budget.
What drives the inference number
Kilo's cost controls guide is candid: model choice, context size, the number of agent steps, retry loops and long unattended runs all add up. A team that defaults everyone to a frontier model for routine edits will pay for it. Auto Model tiers (Frontier, Efficient, Free) exist to route simpler work to cheaper models.
Controls an operator gets
On an organization account you can:
- run one shared credit pool for the team
- set per-user daily spending limits that reset at midnight UTC
- enable auto top-up so work does not stop at a zero balance
- get email alerts when the balance drops below a set amount
- set spend alerts for a rolling window or for sudden hourly spikes
- buy Kilo Pass at the organization level and pool its credits
One trap: personal and organization credits are separate. A developer who tops up a personal account will still hit a low balance warning while working in the team environment.
Visibility across tools
Kilo's Usage Details dashboard shows spend, requests, tokens and active users for Kilo Gateway traffic. It does not see Claude Code, Codex or Cursor, so a developer who also runs those has to check each one separately.
That is where UsageAtlas fits. Each developer installs the free desktop app, connects Kilo Code with an API key and an organization ID, and sees the remaining team balance and Kilo Pass counters on one Limits page with their other tools' limits and resets. It is local-first, needs no account, and does not read prompts or responses. Treat it as a working view, not a billing record; invoices still come from Kilo.
If your team runs Kilo and nothing else, Kilo's own analytics are enough. Add UsageAtlas when the stack is mixed.
Bottom line
Kilo Code is cheap to adopt: the software is free and the platform fee only starts with Teams. Budget for inference, set daily limits and alerts on day one, and give developers one place to see every balance they draw on.