Sample
Sample Workload Report
Illustrative data. These cases, scores, costs, and hardware lines were written for the layout. They are not a customer result and not a measurement.
Planned pricing: from $79. Not for sale. No euro price is listed.
Newsletter
Get the newsletter
Hardware and model notes by email. Joining does not buy this report.
Scope
One tier, shown here with 12 illustrative cases (the tier allows 10–30). Each case is listed as three repeated runs. Scoring, cost, hardware fit, and this page as the export are part of the layout.
| Case | Illustrative pick | Three run scores |
|---|---|---|
| Support reply | Llama 3.2 3B | 3, 3, 4 |
| Refund policy | Llama 3.1 8B | 4, 4, 4 |
| Dutch reply | Gemma 3 12B | 2, 3, 3 |
| Short follow-up | Llama 3.1 8B | 4, 5, 4 |
| Invoice fields | Gemma 3 12B | 3, 3, 2 |
| Tone check | Llama 3.2 3B | 4, 4, 3 |
| Long policy | Gemma 3 12B | 2, 2, 3 |
| Product blurb | Llama 3.1 8B | 5, 4, 4 |
| Escalation | Llama 3.2 3B | 3, 3, 3 |
| Quote wording | Gemma 3 12B | 4, 4, 5 |
| Hours answer | Llama 3.1 8B | 4, 3, 4 |
| Closing line | Llama 3.2 3B | 3, 4, 3 |
Scores are made up so the table has a shape. They are not model output.
Cost
Illustrative API cost for these 12 cases: $1.20. Illustrative rented-GPU note: not priced here. Illustrative local purchase: not a shop listing. None of these figures is an invoice.
Hardware fit
Illustrative sentence: Llama 3.1 8B at Q4 is drawn as a full fit on a 12 GB card, and Gemma 3 12B is drawn as tight. That is layout copy. Tokens per second are not shown, because this sample does not estimate a speed. Use the calculator for an estimate.
Export
This page is the sample export. A later report would use your cases in this shape. Planned pricing.