How much can you earn?
Pick the option that matches your setup. Every tier gets 90% revenue share.
Estimates based on typical usage. Actual earnings depend on volume, pricing, and availability.
Three steps to start earning
Install SpareAI
One command. Detects your hardware and available subscriptions automatically.
npx spareai relay setupChoose what to share
GPU, API key, subscription quota, or all of them. You set the price, you set the limits.
Earn automatically
Requests route to your node. Earnings accumulate in real time. Withdraw anytime in USD or crypto.
Four ways to provide
Pick any combination. Mix and match.
Local GPU
Run open source models on your hardware. Set your price per token.
- Any NVIDIA GPU with 8GB+ VRAM, or Mac Silicon with 16GB+
- Serves Llama, DeepSeek, Qwen, Mistral, and 200+ open models
- OpenAI-compatible endpoint — works with LM Studio, vLLM, Ollama
API Key
Forward requests through your own Anthropic, OpenAI, or Google API key.
- Works with Anthropic, OpenAI, Google Gemini, and more
- Your API usage, your billing — you keep the margin
- Great for unused company API credits or batch plans
Enterprise Credits
Unused AI credits from accelerators or cloud programs? Turn them into cash.
- YC, a16z, Microsoft, AWS startup credits
- Sell at a discount — buyers get cheap API, you get USD
- Marketplace launching Q3 2026
Subscription
Claude Max, ChatGPT Pro, or Gemini Ultra sitting idle? Put it to work.
- Claude Max 20x, ChatGPT Pro, Google AI Ultra
- Built-in anti-ban: rate limiting, quiet hours, randomized timing
- Subscription cost is sunk — every dollar earned is profit
Subscription sharing may violate the provider's terms of service. Use at your own risk. GPU and API key forwarding are fully compliant.
Questions providers ask
It depends on what you share. A Claude Max subscription earns ~$400-600/month. A GPU running open source models earns $50-300/month depending on utilization. Your own API key earnings depend on your unit economics.