RunSnack lets you share your GPU with anyone, anywhere, with a single link. 🔗 Host: run a script → get a link ⚡ Renter: open link → instant GPU terminal in browser 🔒 Sandboxed Docker container, no host access 💸 Zero platform fees — you negotiate directly 👻 Zero-knowledge — P2P WebRTC, no central server sees your data The marketplace lives on Discord: post offers, find GPUs, build reputation.
What's great
remote access (3)free website service (3)user-provided gpus (3)global network of gpus (3)p2p sync (3)
I tried this because I was getting tired of the hassle of renting GPUs through traditional platforms. RunSnack was surprisingly easy to use. A friend ran the setup script, sent me a link, and I was able to access a GPU terminal from my laptop within seconds. No account creation, email verification, credit card, or KYC required.
What I liked most was the P2P approach. There’s no middleman taking a cut, and you can arrange GPU access directly with another person, either by agreeing on a price or simply sharing it with friends. The Docker-based sandboxing also adds an extra layer of protection for the host.
It’s still an early-stage project, though. The Discord community could definitely be more active, and I think having a stronger marketplace/community would make finding available GPUs even easier. That said, I was able to find RunSnack offers through Facebook groups without much trouble.
Overall, if you have an unused GPU or need GPU access for a few hours, this is a very convenient and frictionless option. Definitely worth checking out.
Ratings
Ease of use
Reliability
Value for money
Customization
Report
18 viewsWhat's great
remote access (3)free website service (3)user-provided gpus (3)global network of gpus (3)p2p sync (3)
Tried this because I'm tired of RunPod asking for my soul before letting me use a GPU for 20 minutes — and they didn't even have the Jetsons I needed for my benchmarks. Honestly? It's stupidly simple. A friend ran a script, sent me a link, and I opened it on my laptop. Full GPU terminal in seconds. No account, no email verification, no credit card, no KYC.
The P2P part is what really sold me — there's no platform in the middle. The connection goes directly between machines, so nobody's watching and nobody takes a cut. You negotiate the price directly or just do it for free between friends. Security is handled with a Docker sandbox (read-only, privilege-dropped), so the host machine isn't exposed to whoever connects.
Nonetheless, it's early days — the optional Discord for finding or offering GPUs is still quiet — but I found several RunSnack offers in Facebook groups without much effort. For instant P2P GPU access, this is the most frictionless thing I've tested in ages.
If you have an idle GPU or need one for a few hours, this is free and deserves your upvote.
What needs improvement
The marketplace side — So far, their Discord is too quiet, even when it is easy to find RunSnack offers or provide them in Facebook forums, it would be nice seeing a more active community there.
vs Alternatives
RunpodI am an active user (due to my job conducting benchmarks) of: Vast.ai, TensorDock, SaladCloud, RunPod, Akash, io.net, Nosana, CloudJetson, and more.
I like this fresh gap they cover across the GPU sharing marketplace for zero accounts/fees/setup, direct P2P, and it works with any NVIDIA GPUs — including Jetsons the big clouds don't offer.
Ratings
Ease of use
Reliability
Value for money
Customization
Report
31 viewsWhat's great
remote access (3)free website service (3)user-provided gpus (3)global network of gpus (3)p2p sync (3)
I’ve been using several GPU cloud platforms for benchmark work, and RunSnack caught my attention because of how simple it is to get started. There’s no complicated signup process, KYC, or payment setup just to access a GPU for a short task. Someone can share access with you, and you can connect directly in seconds.
What I like most is the direct P2P approach. You connect straight to the GPU provider without a platform sitting in the middle, and pricing can be arranged directly between both parties. The Docker sandbox also adds a useful layer of isolation for the host.
It’s still a relatively new platform, and the community/marketplace side could definitely be more active. However, I’ve already come across offers outside the main community channels, so I think that will improve as more people discover it.
Overall, it feels like a refreshing alternative for anyone who wants quick, straightforward access to GPUs, especially for short-term tasks or sharing unused GPU resources. The support for different NVIDIA GPUs, including some hardware that is harder to find on larger cloud platforms, is another nice advantage.
What could be improved: The marketplace and community could use more activity. A larger pool of available GPU providers and users in one central place would make it even easier to find resources quickly.
Ratings
Ease of use
Reliability
Value for money
Customization
Report
15 views
NEW — now compatible with DGX Spark (in addition to NVIDIA Jetson (JetPack 6.1) and AMD64) 😊 We keep growing.
UPDATE: added a 30-sec YouTube reel so everyone gets better the idea