Kilo

Kilo is an open-source AI coding agent designed to work across VS Code, JetBrains, CLI, and cloud environments. It provides access to over 500 models with support for custom API keys and local models. The platform is aimed at developers seeking an agentic coding assistant with specialized modes for different software engineering tasks.

• Open-source codebase with MIT licensing for audit and self-hosting
• Access to over 500 models across 60+ providers with zero markup on inference
• Specialized agent modes including Code Mode, Architect Mode, Debug Mode, and Code Reviewer
• Multi-surface support across VS Code, JetBrains IDEs, CLI, and cloud agents
• Auto Model routing that handles strategy selection based on budget and task complexity
• Support for bringing your own provider keys and running local models via Ollama or LM Studio
• Parallel isolated worktrees and session-aware routing

Pros

• Strong model flexibility, allowing users to choose from frontier models, open-weight models, local execution, or custom provider keys
• Consistent workflow across multiple surfaces like IDEs, command-line interfaces, and cloud environments without changing tools
• Transparent sourcing and prompt visibility due to the open-source MIT-licensed nature of the project
• Granular cost and capability control through the Auto Model routing system

Cons

• Platform access, AI inference, and cloud compute are billed separately, requiring users to track multiple components
• Cloud compute usage is billed separately from inference costs
• Advanced team and enterprise governance features require upgrading to paid team or enterprise tiers

Kilo fits individual developers and software engineering teams looking for an open-source, multi-environment AI coding assistant with flexible model choices. It is a poor fit for users who want an entirely closed, proprietary ecosystem or those who do not use VS Code, JetBrains, command-line interfaces, or supported cloud environments.

Kilo operates on a freemium model separating platform access, AI inference, and cloud compute. The Free tier is available at $0 per month for individuals, offering open-source extensions and usage billed separately or run via free/local models and BYOK. The Teams plan costs $15 per user per month with a 14-day free trial, adding team management, usage analytics, shared agent modes, and data privacy controls. The Enterprise tier features custom pricing and includes SSO, audit logs, SLAs, and dedicated support. For AI inference, users can select Auto Free ($0/mo), Kilo Gateway ($0/mo plus usage at exact provider rates), or Kilo Pass subscriptions starting at $19/mo (Starter at $19/mo yielding up to $26.60/mo in credits, Pro at $49/mo yielding up to $68.60/mo in credits, and Expert at $199/mo yielding up to $278.60/mo in credits, with annual options offering 50% bonus credits).

Unlike closed AI coding tools that lock users into specific proprietary models, Kilo is fully open source under an MIT license, letting developers inspect prompts, context windows, and decisions. It provides an agent command center that unifies IDE, CLI, and cloud sessions rather than binding the agent to a single editor. Furthermore, it offers access to over 500 models with zero markup and lets users bring their own keys or run local models.

VS Code, JetBrains (including IntelliJ, PyCharm, and WebStorm), CLI, Cloud Agents, Slack, Ollama, LM Studio, and provider keys for Anthropic, OpenAI, Google, Azure, and AWS Bedrock.