Kilo Code

Kilo Code is an open-source AI coding agent that helps developers write, refactor, debug, and ship production code across the tools they already use. It installs as an extension in VS Code and JetBrains IDEs, runs as a CLI agent, and extends into cloud agents and code review workflows, so the same agent follows you across every coding surface. One of its defining features is model flexibility: Kilo Code connects to more than 500 models from over 60 providers, including frontier models like Claude Opus 4.8, GPT-5.5, and Gemini 3.5, open-weight models, and free experimental options. Users can bring their own API keys, use local models, route through Kilo Gateway, or subscribe to Kilo Pass for bonus credits, and the platform advertises exact provider rates with zero markup. The agent ships with specialized modes: Code Mode for writing and refactoring, Architect Mode for planning complex features before code is written, Debug Mode for tracing errors and suggesting fixes, and cloud agent and code review capabilities for team workflows. Auto Model routing classifies each session and picks an Efficient, Frontier, or Free model strategy so developers can balance capability against cost without micromanaging model selection. Kilo Code is fully open source under the Kilo-Org organization on GitHub, which means transparent pricing, community contributions, and no vendor lock-in. The project positions itself against subscription coding assistants by emphasizing pay-as-you-go flexibility and the freedom to choose any model or provider. It is trusted by teams that want one agent everywhere they code, from solo developers to enterprises that need governance features like model and provider restrictions available on the Enterprise tier. The project's design philosophy centers on user control: there are no artificial ceilings on model choice, no requirement to buy bundled tokens, and no data lock-in, since the agent works with the models and providers you already trust. Kilo Code also ships an Auto Model system that classifies each session in real time and routes work to Efficient, Frontier, or Free model strategies, which is particularly valuable for teams that want frontier-level results on complex tasks while keeping routine requests on cheaper models. The open-source nature of the project means its roadmap is visible and community-driven, with regular releases covering new models, IDE improvements, and agent capabilities. Developers who prefer privacy can run local models entirely offline, while teams can centralize usage through Kilo Gateway for billing, logging, and governance. The product's blog tracks industry shifts such as metered pricing and model availability, positioning Kilo Code as an informed alternative to subscription-based assistants.