# Kilo Code — Open source AI coding agent for VS Code, JetBrains, CLI, and Cloud with 500+ models and no markup

> Source: The Agents Index — https://theagentsindex.com/kilo-code (structured, researched, re-verified)
> Facts last verified: 2026-08-18

Kilo Code is an open source AI coding agent available as a VS Code and JetBrains extension, a CLI, and cloud agents, built for developers who want an AI pair programmer without being locked into one model provider. It is MIT-licensed, so anyone can fork, modify, or self-host it. The project reports over 3 million users and more than 40 trillion tokens processed, and was labeled #1 Open Source Product of the Month. It serves individual developers, teams, and enterprises needing code generation, review, and debugging support across 500+ AI models.

| Fact | Value |
| --- | --- |
| Website | https://kilo.ai |
| Pricing | Platform: Free (Individual) / $15 per user per month (Teams) / Custom (Enterprise); AI inference: provider rates via BYOK or Kilo Pass credits from $19/mo (Starter) to $199/mo (Expert), 5% processing fee on credit purchases |
| API | Yes |
| Not for | Teams that need one predictable monthly bill: Kilo's total cost is a platform plan plus separate, usage-driven inference spend, so it takes more budgeting than a flat-fee competitor. |

## Pricing

| Tier | Price |
| --- | --- |
| Individual (platform) | Free |
| Teams (platform) | $15/user |
| Enterprise (platform) | Custom |
| Kilo Gateway (inference, pay-as-you-go) | $0 + provider rates / month |
| Kilo Pass Starter (inference credits) | $19 / month |
| Kilo Pass Pro (inference credits) | $49 / month |
| Kilo Pass Expert (inference credits) | $199 / month |

## Verdict

Kilo Code positions itself as the open source alternative to closed AI coding assistants: fork it, self-host it, or use Kilo's hosted Gateway to reach 500+ models at provider rates. The mode system (Code, Architect, Debug, Ask, Custom) plus Codebase Indexing and Memory bank cover the core coding-agent workflow, and Cloud Agents and Code Reviews extend it into PR review and longer-running tasks. Pricing is genuinely two-dimensional — a platform plan (Free or Teams at $15/user/month) plus separate AI-inference costs — and independent reviews report that real-world inference spend can run well past the advertised entry price, with one account's credit balance draining unexpectedly. For teams that value auditable, self-hostable code over polish, it's a credible pick; for anyone needing one predictable monthly bill, budget for inference separately from the platform plan.

## How it works

1. Install the VS Code extension via `code --install-extension kilocode.kilo-code`, or the CLI via `npm install -g @kilocode/cli`.
2. Pick an agent mode (Code, Architect, Debug, Ask, or Custom) for the task at hand.
3. Let Auto Model route the request to an Efficient, Frontier, or Free model based on budget, or supply your own API key.
4. Rely on Codebase Indexing and Memory bank so the agent finds relevant files and remembers prior context across sessions.

## Who it's for

- Individual developers who want an AI coding assistant without vendor lock-in
- Teams that want automated PR review paired with in-editor coding help
- Enterprises that need SSO, audit logs, and an SLA around an AI coding tool

## Strengths and weaknesses

- ✓ Open source and MIT-licensed, letting teams audit, fork, or self-host instead of depending on a closed vendor
- ✓ Kilo Gateway charges exact provider rates with zero markup across 500+ models and 60+ providers, so switching models doesn't inflate cost
- ✓ Combines coding, debugging, automated PR review, and vulnerability scanning agents under one product instead of separate tools
- ✗ Pricing runs on two separate axes (a platform plan, plus optional AI-inference credits), so there is no single price tag; an independent cost review estimated $1,000-3,000+/year for moderate real-world use once inference is added
- ✗ A 5% processing fee applies to Kilo Pass credit purchases on top of the subscription cost
- ✗ Independent cost reviews report unpredictable credit burn — one flagged a $5 free credit exhausted in a single day, another an account balance dropping from $74.50 to near zero without active use
- ⚠ Pricing is two-axis: a platform plan (Free / Teams $15 per user per month / Enterprise custom) plus separate, optional AI-inference costs (provider rates via BYOK, or bundled Kilo Pass credits from $19-$199/month) — there is no single number to quote without picking both
- ⚠ Kilo Console, a browser-based console for managing the CLI, is still in Beta
- ⚠ Most feature and usage claims come from the vendor's own site, with independent cost-focused reviews as the main outside corroboration found

## Key features

- **Specialized agent modes** — Code, Architect, Debug, Ask, and Custom modes let a developer switch between writing, planning, debugging, and Q&A without leaving the tool.
- **Auto Model routing** — Automatically routes requests between Efficient, Frontier, and Free model strategies based on budget and task complexity.
- **Codebase Indexing** — Builds a local semantic index across the codebase so the agent finds relevant code without the developer naming files.
- **Memory bank** — Keeps its own records of prior work and user preferences across sessions instead of requiring the app to be re-explained each time.
- **Kilo Gateway** — A single API endpoint routing to 500+ models across 60+ providers at exact provider rates, plus streaming, BYOK, and usage tracking.
- **Automatic failure recovery** — Detects errors, reruns test suites, and pauses and resumes automatically if network connectivity drops mid-task.
- **Security Agent** — Runs AI-powered vulnerability analysis, validating exploitability through sandbox testing before proposing automated fixes.

## Use cases

- **In-editor pair programming** — A developer switches between Code, Architect, and Debug modes inside VS Code or JetBrains to write and reason through a feature without leaving the IDE.
- **Automated pull request review** — The Code Reviews agent analyzes an incoming PR, flags bugs, and suggests improvements before a human reviewer looks at it.
- **Long-running cloud tasks** — A team offloads a multi-step refactor to Cloud Agents so it runs remotely instead of tying up a developer's laptop.
- **Model-agnostic development** — A team routes the same coding task through Kilo Gateway to compare results across Claude Opus 4.8, GPT-5.5, and Gemini 3.5 without switching tools.
- **Security auditing** — The Security Agent sandbox-tests a flagged vulnerability to confirm it's exploitable, then proposes a fix before it ships.

## Integrations

VS Code · JetBrains (IntelliJ, PyCharm, WebStorm) · CLI · Cloud Agents · Slack · Code Reviewer · Kilo Gateway · Gastown · KiloClaw

## Sources

- https://kilo.ai
- https://kilo.ai/pricing/kilo-pass
- https://kilo.ai/api/impact/redirect?target=https%3A%2F%2Fapp.kilo.ai%2Fusers%2Fsign_in
- https://kilo.ai/features
- https://kilo.ai/pricing
- https://www.bodegaone.ai/blog/real-cost-of-kilo-code
- https://kilo.ai/docs
- https://aiagentsquare.com/agents/kilo-code