Skip to main content

Get Claude Code for Free

Use NVIDIA's free AI models as a replacement for paid Claude subscriptions. No credit card required.

15-minute setup | No credit card | Open source

What You Get

Free Claude Code access

Run the full Claude Code CLI without a paid Anthropic subscription. Use powerful AI coding assistance for free.

How It Works

NVIDIA NIM + LiteLLM

NVIDIA provides free AI models. LiteLLM translates requests so Claude Code works normally with your free NVIDIA key.

What You Need

Just 3 things

  • • Docker Desktop
  • • Claude Code CLI
  • • Free NVIDIA account

What is Claude Code?

AI that writes code

Claude Code is an AI coding assistant that can edit files, run terminal commands, and help with complex development tasks directly in your terminal.

What is NVIDIA NIM?

Free AI inference API

NVIDIA NIM provides free access to powerful open-source models like Gemma, Nemotron, and Mistral. No credit card required.

Why LiteLLM?

API translation layer

LiteLLM translates Claude Code's Anthropic API format to OpenAI format that NVIDIA NIM understands. drop_params: true handles compatibility.

Architecture

Claude Code CLI → LiteLLM Proxy → NVIDIA NIM API

Claude Code sends Anthropic-format requests. LiteLLM translates them into OpenAI format and forwards them with your NVIDIA key. Responses are translated back so Claude Code works normally.

Client

Claude Code

ANTHROPIC_BASE_URL and model slot envs

Proxy

LiteLLM

drop_params: true and model mapping

Inference

NVIDIA NIM

gemma, nemotron, mistral endpoints

Requirements

What you need before starting

  • Docker Desktop installed and running
  • Claude Code CLI installed (brew or npm)
  • Free NVIDIA account at build.nvidia.com
  • macOS 13+, Ubuntu 20.04+, or Windows 10+ with WSL

15-minute target

Install CLI, run LiteLLM Docker, add alias, then launch with claude-nim.

No paid Anthropic plan

Use NVIDIA free-tier APIs and keep your key secure through env variables.

Three model slots

Map Sonnet, Opus, and Haiku to optimized NVIDIA models for task fit.