Verification workload
The TJ
Benchmark.
Every card that wants to run creator jobs proves itself the same way: one fixed image fine-tune, timed and measured. It is adapted from NVIDIA's own open-source example, the one that teaches Stable Diffusion XL to draw Toy Jensen. That example is where this project got its name.
Where this comes from.
NVIDIA publishes an AI Workbench example project that fine-tunes SDXL on a handful of photos of their toy mascot. It is public, it is Apache 2.0, and it is a real, repeatable GPU workload. We took it as the yardstick.
NVIDIA AI Workbench: SDXL Customization. Fine-tunes Stable Diffusion XL with DreamBooth so it learns one specific subject. NVIDIA's own example subject is Toy Jensen, the toy figure of their founder that they use as a demo character.
Facts on this page are read from the repository's own README. Verified 10 Sep 2026.
What the upstream project specifies
| Field | Value |
|---|---|
| Base model | Stable Diffusion XL |
| Method | DreamBooth fine-tune |
| Example subject | Toy Jensen |
| Notebook | FineTuning-SDXL.ipynb |
| GPU | Any NVIDIA training GPU |
| Tested on | 1× NVIDIA A100-80GB |
| Storage | 40 GB |
| Operating system | Ubuntu 22.04 |
| Licence | Apache 2.0 |
What we added on top.
The upstream notebook proves a model can learn a subject. We needed it to prove a machine is real, present and as fast as it claims. Four additions do that.
Same run every time
One reference set, one seed, one step count, one resolution. Two cards running it are doing identical work, so the clock is the only variable.
Measured while it runs
The provider app samples NVIDIA DCGM throughout: utilisation, memory in use, clocks, temperature and energy drawn.
A receipt, not a claim
Card, start, end, the samples, the input hash and the output hash are signed by the provider's key. A screenshot cannot fake it.
Written to the chain
The receipt hash goes on Robinhood Chain. The score on a card's passport points at a block anyone can read.
What a run produces receipt fields
The score is the run's wall-clock time normalised against the reference card, then capped at 100. Faster than reference scores above 50, slower below.
Reference times modelled · fills in from real runs in Phase 2
One TJ Benchmark run costs 0.6 hours whichever card takes it, so a slow card earns less per hour of wall time. That is the point. All cards →
Run it yourself
You do not need a GPU to run the benchmark. Spend hours and a registered card runs it for you, and you get the receipt.
Buy hours
1.20 USDG each, paid in your own wallet on Robinhood Chain.
Pick the workload
In the Studio, choose TJ Benchmark fine-tune and how many runs.
Redeem 0.6 h
The job queues for a verified card. Nothing is spent until it is accepted.
Read the receipt
Timing, samples and hashes land in your receipts, and on the card's passport.
Licence, credit and what this is not
What we use. The TJ Benchmark is adapted from NVIDIA's AI Workbench SDXL Customization example project, which NVIDIA publishes under the Apache License 2.0. That licence covers the example's code. We keep the notebook's training procedure and add the fixed parameters, the DCGM sampling, the signed receipt and the onchain anchor described above.
What the licence does not cover. An open-source licence on code says nothing about names, logos or a person's likeness. So this site uses no NVIDIA logo, ships no NVIDIA artwork, and generates no images of any real person as part of the benchmark. The mascot on our home page is our own commissioned figure.
Not affiliated. TOY JENSEN is an independent project. It is not affiliated with, sponsored by, endorsed by or connected to NVIDIA Corporation. NVIDIA, Toy Jensen and related names are the property of their respective owners, and are used here only to say plainly which public project our benchmark is built on.
Read it yourself. The repository ↗ · Apache License 2.0 ↗ · Our open-source page in the docs