Run Claude Code with Google Gemini for Free: A Complete Step‑by‑Step Guide

This guide explains why switching Claude Code to Google Gemini saves money, speeds up responses, and offers longer context, then walks you through obtaining a Gemini API key, configuring Claude Code via a proxy or MCP, verifying the setup, and handling common issues.

Old Meng AI Explorer
Old Meng AI Explorer
Old Meng AI Explorer
Run Claude Code with Google Gemini for Free: A Complete Step‑by‑Step Guide

Why use Gemini with Claude Code

Gemini provides a free tier that is significantly cheaper than Claude, faster response times, and a context window of up to 1 million tokens (about five times larger than Claude).

Obtain a Gemini API key

Access Google AI Studio

Open Google AI Studio (https://aistudio.google.com) and sign in with a Google account.

Create the API key

In the left‑hand menu click Get API key , then Create API key → Create API key in new project . Copy the generated key immediately; it is shown only once.

Google AI Studio Get API key screen
Google AI Studio Get API key screen
Google AI Studio API Keys management screen
Google AI Studio API Keys management screen
AIzaXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

Free quota

Gemini 2.5 Pro – 5 requests/min, 100 requests/day – suited for complex reasoning.

Gemini 2.5 Flash – 10 requests/min, 250 requests/day – ideal for daily development.

Gemini 2.5 Flash‑Lite – 15 requests/min, 1000 requests/day – best for high‑frequency calls.

The quota resets at midnight Pacific Time.

Configure Claude Code to use a third‑party endpoint

Claude Code expects Anthropic‑compatible API calls, while Gemini uses Google’s format. Two integration approaches are available:

Use a proxy service that translates Claude‑style requests to Gemini (e.g., laozhang.ai, OpenRouter).

Integrate Gemini directly via the Model Context Protocol (MCP).

Prepare configuration files

Create a dummy primary API key file to bypass Claude’s validation:

# macOS / Linux
mkdir -p ~/.claude
echo '{"primaryApiKey": "any-string-is-ok-here"}' > ~/.claude/config.json

# Windows PowerShell
$path = "$HOME\.claude"
New-Item -ItemType Directory -Path $path -Force | Out-Null
Set-Content -Path "$path\config.json" -Value '{"primaryApiKey": "any-string-is-ok-here"}'

Skip the onboarding flow:

# macOS / Linux
echo '{"hasCompletedOnboarding": true}' > ~/.claude.json

# Windows PowerShell
Set-Content -Path "$HOME\.claude.json" -Value '{"hasCompletedOnboarding": true}'

Proxy service configuration (example: laozhang.ai)

Obtain an API key from laozhang.ai after registration.

Edit ~/.claude/settings.json (create if missing) with the following content:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.laozhang.ai/v1",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_LAOZHANG_API_KEY",
    "CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC": 1
  }
}

Set the environment variables for the current shell:

# Bash (macOS / Linux
export ANTHROPIC_BASE_URL="https://api.laozhang.ai/v1"
export ANTHROPIC_AUTH_TOKEN="YOUR_LAOZHANG_API_KEY"

# PowerShell (Windows)
$env:ANTHROPIC_BASE_URL = "https://api.laozhang.ai/v1"
$env:ANTHROPIC_AUTH_TOKEN = "YOUR_LAOZHANG_API_KEY"

Verify the setup

Open a new terminal (environment variables are only loaded in new shells) and run: claude After the welcome screen, type /status to view the active model and endpoint, for example:

Model: gemini-2.5-flash
Endpoint: https://api.laozhang.ai/v1

Advanced integration via MCP

Install the MCP server

# Clone the repository
git clone https://github.com/Raydius/gemini-for-claude-mcp.git
cd gemini-for-claude-mcp

# Install dependencies
npm install
npm run build

Add the MCP plugin to Claude Code

claude mcp add gemini \
  -e GEMINI_API_KEY=YOUR_GEMINI_API_KEY \
  -e GEMINI_DEFAULT_MODEL=gemini-2.5-flash \
  -- node /absolute/path/gemini-for-claude-mcp/dist/app.js

Using the MCP plugin

Invoke Gemini through Claude Code with natural‑language prompts, e.g.:

Help me explain Rust's ownership model with Gemini.

Claude forwards the request to Gemini and returns the result.

Model selection guidance

Simple scripts / rapid prototyping – Gemini 2.5 Flash‑Lite

Daily development & debugging – Gemini 2.5 Flash

Complex tasks & code review – Gemini 2.5 Pro

Ultra‑long context processing – Gemini 1.5 Pro

Common issues

429 – quota exceeded

Reduce request frequency, switch to a model with a larger quota (e.g., Flash‑Lite), or wait for the daily reset.

Access from Mainland China

Google AI Studio is blocked; use a VPN or a proxy service such as laozhang.ai to reach Gemini.

Configuration changes not taking effect

Restart the terminal so environment variables are loaded.

Ensure ~/.claude/settings.json exists at the correct location.

Verify the API key is correct.

Check the base URL spelling.

Free quota exhausted

Wait for the next day’s reset.

Upgrade to a paid tier if additional calls are required.

AI Coding AssistantProxy serviceGoogle GeminiClaude Codesetup guideMCP integrationFree API
Old Meng AI Explorer
Written by

Old Meng AI Explorer

Tracking global AI developments 24/7, focusing on large model iterations, commercial applications, and tech ethics. We break down hardcore technology into plain language, providing fresh news, in-depth analysis, and practical insights for professionals and enthusiasts.

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.