Kilo Code — Free AI Coding Setup Guide

A complete guide to installing and configuring Kilo Code (an AI coding agent for VS Code) using the Google Gemini free tier as your primary model and OpenRouter free models as a backup/variety layer — no cost, no credit card required.

⚠️ Important: As of the last verification date, Google Gemini has no free model on OpenRouter. Gemini's real free tier only exists when you get an API key directly from Google AI Studio, not through OpenRouter. This repo is written to reflect that reality — see Architecture and Limitations below.

---

📚 Table of Contents

• Introduction
• Architecture
• Quick Start
• Full Setup
• Recommended Models
• Advanced Configuration
• Best Practices for AI Coding
• Security
• Limitations
• Troubleshooting
• FAQ
• Contributing
• License

---

🎯 Introduction

Kilo Code is an open-source AI coding agent that runs inside VS Code (also available for JetBrains and CLI), supporting more than 30 different model providers. This repo shows you how to connect Kilo Code to genuinely free, verified models so you can code with AI assistance at no cost, within free-tier limits.

This repo is useful for:

• Students / individual developers who want to try an AI coding agent before paying for one.
• Newcomers to VS Code extensions who need a clear, step-by-step guide with screenshots.
• Anyone who wants to know where the real free tier actually lives, instead of trusting outdated guides.

---

🏗️ Architecture

VS Code
   │
   ▼
Kilo Code (extension)
   │
   ├── Provider: Google Gemini (direct, via Google AI Studio)
   │     └── Genuine free tier — recommended as your primary model
   │
   └── Provider: OpenRouter (optional, backup)
         └── The list of :free models changes constantly — use it for
             variety, NOT to get Gemini for free (Gemini isn't free on OpenRouter)

Kilo Code doesn't have its own models — it needs a provider to supply inference. There are two independent paths:

Path	Provider in Kilo Code	Where to get a key	Gemini free tier?
A (recommended)	Google Gemini	Google AI Studio	✅ Yes
B (backup)	OpenRouter	openrouter.ai/keys	❌ No (but other free models are available)

---

⚡ Quick Start

# 1. Install the Kilo Code CLI (optional — or install the extension from the VS Code Marketplace)
npm install -g @kilocode/cli
 
# 2. Get a free Gemini API key at:
# https://aistudio.google.com/apikey
 
# 3. In VS Code:
# Kilo Code panel → gear icon (Settings) → API Provider → "Google Gemini"
# → paste your API key → pick a model → start coding

---

🔧 Full Setup

Step 1 — Install VS Code + Kilo Code

1. Install Visual Studio Code if you don't have it yet.
2. Open the VS Code Marketplace, search for Kilo Code, click Install.
3. Reload VS Code once installation finishes. ✅ Expected result: the Kilo Code icon appears in the left sidebar.

Step 2 — Get a Gemini API Key (free)

1. Go to Google AI Studio.
2. Sign in with your Google account.
3. Click Create API key.
4. Copy the generated key — this is an Auth key, distinct from the old Standard key (Google stopped accepting unrestricted Standard keys starting mid-2026).

Step 3 — Configure Kilo Code to use Gemini

1. In the Kilo Code panel, click the ⚙️ (Settings) icon.
2. Go to the Providers tab.
3. Under API Provider, select Google Gemini.
4. Paste your key into the Gemini API Key field.
5. Under Model, choose the Gemini model you want (see the recommended table below).
6. Save your settings. 🔍 Verification: type a simple prompt in the Kilo Code chat (e.g. "write a Fibonacci function in Python"). If you get a response, the setup is working.

Step 4 (optional) — Add OpenRouter as a backup provider

1. Go to openrouter.ai, sign in with Google or GitHub.
2. Go to openrouter.ai/keys and create a new API key.
3. In Kilo Code Settings → Providers, add OpenRouter, paste your key.
4. Go to openrouter.ai/models and filter by :free to see the current list of free models — this list changes weekly, so don't hardcode it.

---

🤖 Recommended Models

Last verified: 2026-08. The free model lineup changes frequently — always double-check the official source before relying on this long-term.

Model	Provider	Free?	Good for coding	Context	Notes
Gemini 2.5 Flash	Google AI Studio (direct)	✅	Good, fast	Large	Recommended default
Gemini 2.5 Flash-Lite	Google AI Studio (direct)	✅	Decent, very fast	Large	Use for speed, simple tasks
Gemini 2.5 Pro	Google AI Studio (direct)	✅ (lower limits)	Excellent, deep reasoning	Largest	Use for complex tasks, to conserve quota
OpenRouter :free models	OpenRouter	✅ (not Gemini)	Varies by model	Varies	Check the live list at openrouter.ai/models

---

⚙️ Advanced Configuration

• Kilo Code's config file: kilo.json — can be edited manually to add multiple providers and set a model per mode (Architect/Code/Debug).
• With OpenRouter, you can enable the "middle-out" transform to handle prompts that exceed a model's context window, via options.transforms in kilo.json.
• You can configure several providers in parallel and switch between models manually mid-session.

---

✅ Best Practices for AI Coding

• Break tasks into small steps: give specific, incremental instructions instead of one large, vague prompt.
• Let the agent read the project first: ask Kilo Code to scan the relevant folder/file structure before making changes.
• Use fast models (Flash) for repetitive work, and stronger models (Pro) for architecture decisions or complex debugging.
• Always review the diff before accepting changes — don't let the agent overwrite multiple files at once without checking.
• Scope edits explicitly: specify exactly which files/folders the agent is allowed to touch.

---

🔒 Security

• Never commit API keys to Git.
• Always use a .env file and add .env to .gitignore.

GEMINI_API_KEY=your_api_key_here
OPENROUTER_API_KEY=your_api_key_here

• If you suspect a key has leaked: go to the Google AI Studio or OpenRouter dashboard → revoke the key immediately → create a new one.
• Don't use Google's old Standard API keys (no longer accepted) — always use the newer Auth key type.

---

⚠️ Limitations

• "Free" here means a free tier with limits (per-minute/per-day rate limits), not "free forever."
• Free tiers and model availability can change at any time, at the provider's discretion (Google, OpenRouter).
• Gemini does not have a free tier on OpenRouter as of this writing — it's only free when you get a key directly from Google AI Studio.
• Always check the official documentation before relying on any specific number in this repo.

---

🛠️ Troubleshooting

Error	Cause	Fix
401 Unauthorized	Wrong API key, or using an old Standard key	Create a new Auth key in Google AI Studio
429 Too Many Requests	Exceeded free-tier rate limit	Wait for the limit to reset (per minute/day), or reduce request frequency
Model unavailable / removed	The free model was pulled from OpenRouter	Check openrouter.ai/models again and pick another one
Kilo Code not responding	Provider isn't configured correctly	Go to Settings → Providers and double-check the key and selected model
"Free Gemini on OpenRouter" doesn't work	Gemini currently has no free tier on OpenRouter	Get a key directly from Google AI Studio instead

---

❓ FAQ

Do I need a credit card for the free tier? No — both Google AI Studio and OpenRouter let you start with just a Google/GitHub account.

Can I use the free tier for a production project? Not recommended — rate limits are low, there's no uptime guarantee, and your data may be used to improve the model (depending on the provider's policy).

Is the free model list in this repo always accurate? No — the model table is only accurate as of its "last verified" date. Always check the official source before relying on it long-term.

---

🤝 Contributing

This repo is maintained as a living guide. Contributions are welcome, especially:

• Updating models/rate limits when they change.
• Adding newly discovered errors/fixes.
• Improving instructions or adding screenshots. See CONTRIBUTING.md for details.

---

📄 License

Released under the MIT license.