Open source · $0

Free Inference Infrastructure

Use the gateway
4,979,948,851
tokens served through free tiers · still $0.00
For coding agents
For your own tools
Paid for by nobody
$ curl -sSL https://api.free-ride.xyz/ridex.sh | sh
ridex, the agent

A coding agent that runs on $0.

One command installs the agent and its gateway, supervises the daemon, and drops you into a session. The status line reads $0 because it is.

+Supervised daemon

launchd and systemd restart the gateway if it ever dies

+Fallback ladder

up to four healthy candidates lined up behind every request

+Self-diagnosing

ships a skill that understands and debugs its own gateway

+No login

first run asks for one free API key, and that's the whole setup

ridex
ridex v0.1.3 · Run /help for commands
hello world
Hello! What can I help you with today?
2s (↑1 ↓11) · $0.00
freeride/coding · $0 · no login
gateway.log
[12:41:03.221]POST/v1/chat/completions referer=continue.dev provider=groq status=200
[12:41:03.301]POST/v1/messages referer=claude-cli provider=openrouter status=tool_use
[12:41:03.487]POST/v1/chat/completions referer=aider provider=openrouter status=stream
[12:41:03.903]GET/v1/models referer=cursor provider=cache status=200
[12:41:04.118]POST/v1/chat/completions referer=hermes provider=nvidia_nim status=200
[12:41:04.401]POST/v1/messages referer=claude-cli provider=openrouter status=stream
[12:41:04.516]POST/v1/chat/completions referer=openclaw provider=huggingface status=cool
[12:41:04.882]POST/v1/chat/completions referer=continue.dev provider=cerebras status=200
[12:41:05.104]GET/health referer=cli provider=local status=200
[12:41:05.433]POST/v1/chat/completions referer=aider provider=groq status=200
[12:41:05.701]POST/v1/chat/completions referer=openclaw provider=openrouter status=429
[12:41:05.903]POST/v1/chat/completions referer=openclaw provider=nvidia_nim status=200
[12:41:03.221]POST/v1/chat/completions referer=continue.dev provider=groq status=200
[12:41:03.301]POST/v1/messages referer=claude-cli provider=openrouter status=tool_use
[12:41:03.487]POST/v1/chat/completions referer=aider provider=openrouter status=stream
[12:41:03.903]GET/v1/models referer=cursor provider=cache status=200
[12:41:04.118]POST/v1/chat/completions referer=hermes provider=nvidia_nim status=200
[12:41:04.401]POST/v1/messages referer=claude-cli provider=openrouter status=stream
[12:41:04.516]POST/v1/chat/completions referer=openclaw provider=huggingface status=cool
[12:41:04.882]POST/v1/chat/completions referer=continue.dev provider=cerebras status=200
[12:41:05.104]GET/health referer=cli provider=local status=200
[12:41:05.433]POST/v1/chat/completions referer=aider provider=groq status=200
[12:41:05.701]POST/v1/chat/completions referer=openclaw provider=openrouter status=429
[12:41:05.903]POST/v1/chat/completions referer=openclaw provider=nvidia_nim status=200
FreeRide, the gateway

One endpoint for every free provider.

Already have an agent? Keep it. The gateway speaks your SDK's dialect and does the provider juggling behind one URL.

+Four dialects

OpenAI, Anthropic, Gemini, and Responses on one port

+Key rotation

per-key cooldowns for rate limits, auth errors, and quotas

+Smart routing

ask for freeride/coding, fast, quality, or free

+Honest errors

structured 503s when every candidate is cooling, so your retry logic stays in charge

$ export OPENAI_API_BASE=http://localhost:11343/v1

Recently shipped

Run your agent on $0.

Star on GitHub