Kilo Code
Both form factors at once; open, free clients, $15 per seat for a team, inference at the provider's rates with no markup.
Verified: Sep 13last change: Sep 25next price check: Dec 12
Facts with sources
- Acquired by Anaconda on July 15; about 10 trillion tokens a month pass through the platform from 3 million developers. Source (in Russian)
- A 5% fee applies only to topping up credits; there’s no markup on top of the provider’s rates. Source (in Russian)
- Alerts for unusual spend and a per-person cost breakdown showed up where billing runs through the key. Source (in Russian)
Strengths
- VS Code, JetBrains and CLI extensions are free and MIT-licensed, and work with any model.
- Kilo Gateway routes to 500+ models from 60+ providers at provider rates with no markup; your own key or a local model costs $0.
- Slack bot: mention @Kilo in a thread and it opens a branch and submits a pull request, working out which repo you mean.
- Teams is $15 per user per month with analytics, shared modes and central billing; cloud agents are billed per second.
Weaknesses
- GitHub #1557: a balance dropped from $74.50 to $0.03 with no recorded usage, and the team closed the issue as "not planned".
- GitHub #3767: a 340-line file sent the agent into a loop, context grew from 140k to 262k tokens, and a $0.10 review cost $7.59.
- GitHub #1388: the extension silently switched to Kilo's hosted Sonnet and burned credits even for users on their own key.
- Trustpilot 2.6 from 17 reviews, 59% one-star: slow responses, the agent circling one task, and a panel with no tabs.
Example from practice
A week checking five Kilo claims
The reviewer spent a week testing five of Kilo's marketing claims against the docs, the Kilobench leaderboard and open GitHub issues. Bring-your-own-key with no markup and the Slack bot that ships a pull request from a thread both held up. The "$0 a month, no hidden costs" claim fell apart on documented charges: $74.50 gone with no usage, and $7.59 for a review that should have cost $0.10. Final tally: three claims verified, one wrong, one a tie.
source: Bryan Software, Kilo Code: My Honest Take After a Week
What reviewers say
Reviewer opinions on their own tasks and hardware, not measurements. Links go to the videos; transcripts live in docs/raw-data.
- Kilo Code ai Review: The Best Free Alternative to Cursor and Claude Code · Riflo animations, Sep 8In the reviewer's view, Kilo is the best free alternative to Cursor for tinkerers willing to supervise the agent; it is not automatically cheaper and is less polished.
- Kilo Code: My Honest Take After a Week · Bryan Software, Aug 28The reviewer recommends Kilo for model freedom and Slack-native pull requests, but only with hard spending limits; anyone who cannot tolerate unpredictable billing should pick Cursor.
Who it is for
- Team that needs terminal, IDE, and a predictable per-seat bill
Models in this agent
Score is from our manager-task benchmark; models without a score are not measured yet.
What changed
- Kilo Code integrated Enkrypt AI guardrails: admins define policies once and developers keep working as usual →
- Kilo Swarm is on by default: agents and subagents exchange messages on a shared board; the vendor measures a 39% cost cut on the tasks it is built for and inconclusive results elsewhere →
- Kilo Code can import global Claude Code settings once (an experimental feature, off by default) →
- Kilo Code now shows an Enkrypt safety score for models on a 0–100 scale, lower is safer →
- Kilo Code added Auto Model: the shell picks a model per request, the vendor claims about 30% savings on mid-size tasks →
- GPT-6 Astra and Claude Fable 5.1 are available in Kilo Code across VS Code, JetBrains and the CLI →
- Kilo Code is on the AWS Marketplace: $180 a year per user for teams, procurement without a new vendor →
- Kilo Code released a mobile app: agent sessions, review and merging pull requests from a phone
- Kilo Code acquired by Anaconda: about 10 trillion tokens a month pass through the platform
How to brief an agent like this
The four-line brief is the Foundation module: prompt structure, LLM agents and MCP, an AI policy for the team. It starts with the free open module.