What the research costs
The studies published here run on real provider spend. This page sets out what that costs, and what a donation changes.
Jackbench's hosted baseline scored a 153 USD effective sample across 7,255 provider requests. The current bridge run has a 500 USD ceiling. All in, roughly 1,000 USD has gone into the study so far.
Table 1 · Jackbench provider spend
| Item | Amount (USD) | Notes |
|---|---|---|
| Hosted baseline, scored effective sample | 153 | 7,255 provider requests |
| Bridge run (current) | ≤ 500 | running under a 500 USD ceiling |
| Programme to date | ~1,000 | all runs, 2026 |
Figures from sealed per-attempt receipts.
What a donation buys
Attempts. More attempts mean more models in the grid, tighter intervals, and confirmatory gates that can afford to be strict. A donation doesn't buy influence over the results: the analysis plan is registered before the money is spent.
There are no tiers and no perks. If the write-ups are useful to you, this is how more of them happen.