Run open-weight models on hardware you already own, reach them from any device, and tap a distributed network when you need more than your machine can handle. Encrypted end to end, always.
Run models on your own hardware and reach them from any device. End-to-end encrypted, so the server can't read your data. Earn credits while your machine sits idle.
Tap the distributed network for models your hardware can't run. Spend credits to access them, encrypted and anonymized end to end.
macOS · Windows · Linux · Open-weight models · OpenAI-compatible API
Cloud AI keeps getting pricier, and local tools stay chained to one device. You end up choosing between privacy and convenience.
Developers pay for AI tools every month while capable hardware sits idle at home.
You hit the ceiling on hosted models halfway through the workday. Again.
Local tools like Ollama are great, until you step away from the machine running them.
Most platforms can read your prompts and train on them. Spore uses real end-to-end encryption, so we can't.
Your M4 Mac at home could run these models. It just can't help you when you're away from it. Spore keeps the model on your hardware and gives you a private, encrypted way to reach it from anywhere.
One-click setup for macOS, Windows, and Linux. Grab the download from the web app.
Choose the models you want, host them on your hardware, and chat with them from any device.
Let your node process public requests when you're not using it. Credits accumulate on their own.
Need a model your hardware can't run? Spend credits to use it across the network, still encrypted end to end.
On-device inference is powered by llama.cpp.
Turn on community serving and your node processes public inference requests whenever it's idle, earning credits on its own. Or keep it entirely to yourself. Either way, you decide.
Real cryptographic privacy, not a policy promise. Personal requests stay fully private; community requests are encrypted and anonymized, so processing nodes never learn who you are.
A curated set of the best open-weight models. No vendor lock-in, ever.
Get to your models from any device, anywhere, with no third-party software in the middle.
An OpenAI-compatible API. Point your existing code at Spore and it just works, no rewrites.
# OpenAI-compatible: swap your base URL client = OpenAI( base_url="https://api.sporeintel.com/v1", api_key="sk-...", ) client.chat.completions.create(...)
Start free and stay free for personal use. Upgrade only when you want the best rates on the network.
For getting started
For heavy users who want the best rates
Stop renting AI. Run it yourself, reach it from anywhere, and earn from the time your machine would spend idle.