Compared with other ways to give
Folding@home, BOINC projects, giving cash and paid compute marketplaces. Where each is stronger, where SpareTokens fits, and our sources.
Where things stand. SpareTokens is a private pilot with no public worker and no discoveries claimed. The other options below have track records we do not. This page is meant to help you pick the right way to give, and sometimes that is not us.
The short version
- Folding@home and BOINC projects turn spare computer time into science, with decades of experience. Pick them if you have an idle computer and want a proven project.
- Giving cash to a research charity funds people, labs and equipment, which AI credit cannot. Pick it if you want to fund researchers directly.
- Paid compute marketplaces let you buy or sell compute. Pick them if you need compute or want to earn from hardware.
- SpareTokens puts AI model capacity, such as expiring API credit, toward open problems where a machine can check the answer. Pick it if that is the spare thing you have.
Side by side
| Question | SpareTokens | Folding@home | BOINC projects | Cash to a research charity | Paid compute marketplaces |
|---|---|---|---|---|---|
| What you give | AI model capacity: API credit under a cap, or a local model on idle hardware | Computer time on your own machine | Computer time on your own machine | Money | Compute, for payment, or you pay for compute |
| What it goes toward | Open problems with a machine-checkable answer, starting with Erdős problems | Protein and disease simulation | A catalog of projects, such as astronomy, biology and math | Whatever the charity funds, including people and experiments | Whatever the buyer wants to run |
| Cost to you | The API spend you cap, or electricity for a local model | Electricity | Electricity | The amount you give | Varies. You pay to rent, or you earn |
| How results are checked | Code we own re-checks each result before it counts | Published and reviewed by the research groups | Varies by project. Many compare redundant runs of the same work | Through the funded research and its peer review | Varies by network |
| Effort to start | About ten minutes, and the worker is not public yet | Install a client | Install a client and pick projects | A few minutes to donate | Accounts, wallets or a provider setup |
| Track record | Private pilot. No users claimed, no discoveries claimed | Decades of operation and a large published record | Decades of operation across many projects | Depends on the charity | Varies, and independent usage data is thin |
| Credit for you | A username on a leaderboard ranked by verified work, never by spend | Teams and statistics | Credit, teams and statistics | Depends on the charity | Payment |
Where others are better
Folding@home
It is proven. By its own account it passed 280,000 GPUs at the 2020 peak, and it has years of scientific output and an established community. If you have an idle computer and want to help with biomedical research today, it is a sound choice, and you can run it next to SpareTokens. SpareTokens is independent and not affiliated with it.
BOINC projects
BOINC is a mature platform with many projects, so you can pick a cause you care about, and its volunteers have decades of experience with credit, teams and cheating. Its credit systems have also taught hard lessons that we learned from: credit should follow verified work, and communities guard what credit means. If you want a proven project for your idle computer, start there. A 2014 report counted about 245,000 active users across 47 projects.
Giving cash to a research charity
Money funds scientists, equipment and experiments in the real world, which AI credit cannot do. Charities can also choose what to fund with expert review. If you want to back researchers directly, a donation is the better tool, and it can go alongside spare capacity.
Paid compute marketplaces
If you need to buy compute, or want to earn from hardware you own, a marketplace or a cloud provider is the right tool, and SpareTokens does not sell compute. We pay contributors nothing and use no token. Coverage of these networks is mostly overviews, and we found no audited utilization data, so treat any numbers you see with care.
Where SpareTokens fits
- Expiring credit. Prepaid and promo API credit that would otherwise lapse has few other homes. Existing programs give credits to researchers. We found none that lets individuals donate their own unused credit.
- Machine-checkable science. We only hand out tasks where a machine can check the answer, because projects that could not check results lost their communities’ trust. The Equational Theories Project showed what a closed goal and a proof checker can do.
- Caps and honesty. Caps are enforced on your machine, keys never leave it, and we say plainly what has and has not run.
What we do not yet have: public users, results beyond known values, a track record, and tasks beyond math. See the results so far and the roadmap with honest status.
Sources
Research notes dated 20 November 2026. Forum posts and vendor pages are marked as such in our notes and were not used for numbers here. Figures are as reported on the date of each source.
- Folding@home, 2020 in review: its own account of over 280,000 GPUs at the 2020 peak.
- TechCrunch, March 2020: volunteer growth during the COVID surge.
- Nature, 2014: about 245,000 active BOINC users across 47 projects, down from about 350,000 in 2008.
- BOINC paper, 2019: about 700,000 devices, around 560,000 of them GPUs.
- BOINC volunteer poll, 2006: self-selected, ranked fair and prompt credit far above generous credit.
- SETI Institute notice: why SETI@home stopped sending work in 2020.
- GIMPS announcement, 2024: a 41-million-digit prime, confirmed by independent runs.
- Equational Theories Project paper: a crowd of people and tools, with every result checked in Lean.
- CoinGecko overview of decentralized compute: mostly overview coverage, not audited utilization data.
- Crypto.com research on decentralised compute: directional only.
- Anthropic AI for Science program: credits for academic and nonprofit researchers.
- OpenAI Researcher Access program: credits for researchers.