Claude Code (cc) — Free LLM API Config

Point Claude Code at any free OpenAI-compatible backend in 30 seconds.

How It Works

Claude Code reads ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN from the environment. When you set them to a free provider, cc routes API calls through that backend instead of Anthropic's paid API.

Quick Config

Groq (fastest inference, no credit card)

export ANTHROPIC_BASE_URL="https://api.groq.com/openai/v1"
export ANTHROPIC_AUTH_TOKEN="gsk_your_groq_api_key"

Get your key at console.groq.com/keys. Recommended model: llama-3.3-70b-versatile, qwen-3-32b.

NVIDIA NIM (no daily token cap)

export ANTHROPIC_BASE_URL="https://integrate.api.nvidia.com/v1"
export ANTHROPIC_AUTH_TOKEN="nvapi-your_nvidia_key"

Get your key at build.nvidia.com. Recommended model: meta/llama-3.3-70b-instruct, deepseek-ai/deepseek-r1.

OpenRouter (35+ free models, single key)

export ANTHROPIC_BASE_URL="https://openrouter.ai/api/v1"
export ANTHROPIC_AUTH_TOKEN="sk-or-v1-your_openrouter_key"

Get your key at openrouter.ai/keys. Recommended model: openai/gpt-oss-120b:free, google/gemini-2.5-flash:free.

Cloudflare Workers AI (10K requests/day free)

export ANTHROPIC_BASE_URL="https://api.cloudflare.com/client/v4/accounts/YOUR_ACCOUNT_ID/ai/v1"
export ANTHROPIC_AUTH_TOKEN="your_cloudflare_api_token"

SiliconFlow (best latency for China-based users)

export ANTHROPIC_BASE_URL="https://api.siliconflow.cn/v1"
export ANTHROPIC_AUTH_TOKEN="sk-your_siliconflow_key"

Get your key at cloud.siliconflow.cn. Recommended model: Qwen/Qwen3-Coder-480B-A35B-Instruct.

Persistent Config

Add to your shell profile (~/.zshrc or ~/.bashrc):

# Free LLM API backend for Claude Code
export ANTHROPIC_BASE_URL="https://api.groq.com/openai/v1"
export ANTHROPIC_AUTH_TOKEN="gsk_your_key_here"

Caveats

Claude Code expects tools/capabilities that not all free backends support. If you hit errors, try a more capable model (Llama 3.3 70B+ or DeepSeek R1).
The "Claude" model name shown in cc is cosmetic — you're actually using whichever model you configure.
Free tiers have rate limits. Groq: 14,400 RPD. NVIDIA NIM: 40 RPM. OpenRouter: 50 RPD (free tier).

More Configs

Visit freellm.net/config/claude-code/ for the interactive config generator with all 128+ free models.