Free Claude Code.Now on fast local Gemma 4.
Free Claude Code runs the real Claude Code CLI against free models instead of paying per token. Now that Gemma 4 is fast on Apple Silicon, I pointed it there — so you get real Claude Code coding, running fully on your own Mac, free and fast, with nothing leaving your machine. Here's how I wired it in, with a real run.
Gemma 4 got ~90% faster — so local Claude Code is usable
Free Claude Code has always let you run the Claude CLI on a free model. The problem was speed — a slow local model made the coding panel painful. This changed that: Ollama turned on multi-token prediction for Gemma 4 via Apple's MLX engine, so the local model is now fast enough to actually code with. That's what makes free, local Claude Code genuinely usable.
What this is built on.
I wanted Claude Code without the meter.
Before
I love the Claude Code CLI, but every session ran up a bill.
So I'd hesitate before letting it loose on a big task, watching the cost.
Free Claude Code let me point it at a free model — but the free local option was too slow to enjoy.
So I kept paying, or kept it short.
Then Gemma 4 got fast on my Mac — and I pointed Free Claude Code straight at it.
After
Now I run the real Claude Code CLI on a local model, free, at a speed that feels fine.
My private code never leaves my machine.
I stopped watching the meter and just let it work.
Real Claude Code. Zero per-token cost. Fully on your own Mac.
Why listen to me on this.
I run an AI agency with 70+ people where AI handles about 80% of the ops, and a room of operators coding with AI across every kind of business.
No invented quotes. The wins are real and written by the members themselves — agency owners, ecom founders, creators, solo operators across 38 countries. Read them in their own words.
Read the 158-page wins doc →Commit to one free coding session tonight.
Below is exactly how I wired Free Claude Code to local Gemma 4 — one line — plus a real run.
Here's the deal I want to make with you.
Before you sleep tonight, point Free Claude Code at Gemma 4 and run one real coding task on it. Free, fully on your own machine.
Because once you've had the Claude Code experience with no meter running, you start reaching for it constantly — and that changes how much you build.
Be one of the people who runs it today.
Commit to the transition. One free, local coding session tonight.
Real Claude Code — pointed at a free local model.
Here's the trick. Free Claude Code runs a tiny local proxy (fcc-server) that speaks Claude's own API but routes your requests to whatever model you choose — OpenRouter, Kimi, or a local one. Point that at Ollama's fast Gemma 4, and the actual Claude Code CLI runs on a free local model without knowing the difference.
One line to switch it to Gemma 4.
Free Claude Code reads its model from ~/.fcc/.env. Point it at the fast MLX Gemma 4 (Ollama's a supported local provider), and restart the proxy:
# in ~/.fcc/.env
MODEL="ollama/gemma4-mlx" # fast MLX+MTP Gemma 4, local + free
# (Ollama is already a provider — OLLAMA_BASE_URL=http://localhost:11434)
# then restart the proxy
fcc-serverThat's it. Now the Claude Code CLI — with ANTHROPIC_BASE_URL pointed at the local proxy — runs every request on Gemma 4. No API key charged, nothing leaves your Mac.
I asked Free Claude Code for a prime-number function.
Here's an actual response from the proxy — note the model tag at the top: it really is running on local Gemma 4, and the code is correct.
What changes the moment you switch.
Free Claude Code is already wired inside the Agent OS.
You can do the config above yourself — or open the Agent OS, where Free Claude Code on local Gemma 4 is already set up, next to everything else, sharing one memory.
You're not buying a tool. You're getting the whole operating system I run a seven-figure business on.
Get the Agent OS →No — this guide is a perfect example of why not. Free Claude Code on local Gemma 4 means the real Claude Code experience costs $0 per token and runs on your own machine. The Agent OS runs the everyday 90% on a free local model, slots in free APIs, and for the frontier stuff drives the CLIs you already pay for (your Claude subscription already includes the Claude CLI — so you're not paying twice).
It's a layer on top of what you already own, not a new meter. Plus the Boardroom has full token-optimisation tutorials.
How most people run Claude Code. And how I do it now.
- Every Claude Code session costs tokens
- You hesitate before big tasks
- Your code goes to a cloud API
- The free local option was too slow
- So you kept paying, or kept it short
- The real Claude Code CLI, free per token
- Let it loose — nothing's ticking
- Your code never leaves your Mac
- Fast local Gemma 4 makes it usable
- One line in a config file to switch
What's holding you back.
Wrong: "Free Claude Code must be a watered-down clone."
Right: It's the actual Claude Code CLI — the real tool — just pointed at a different model through a local proxy. Same commands, same experience, no per-token bill.
Wrong: "A local model can't write real code."
Right: The run above is real — Gemma 4 wrote a correct prime-number function, tagged as the local model. For everyday coding help, a fast local Gemma 4 is genuinely useful.
Wrong: "Setting up a proxy sounds hard."
Right: It's one line in a config file and one command to restart. And inside the Agent OS it's already done — you just start coding.
158 pages of members already coding with this stack — real businesses, real wins, in their own words.
Read the 158-page wins doc →The Claude Code experience, with no meter.
Free Claude Code on fast local Gemma 4 is the real coding tool, running for free on your own machine. That's the kind of thing that quietly changes how much you build in a week.
Inside the AI Profit Boardroom you get the full Agent OS — Free Claude Code on Gemma 4 already wired, the fast local engine, the Claude Workspace where every build is saved, Agent Kanban, every CLI you already pay for in one dashboard, the AI Mastermind, memory that knows your codebase, a 30-day roadmap, daily tutorials, coaching calls, and 3,600+ founders across 38 countries building alongside you. Every new update gets folded in the week it lands.
It's the operating system I run a seven-figure business on. You get the whole thing.
Get the Agent OS →Point Free Claude Code at Gemma 4 and run one task tonight. I'll see you in the next one.
