A benchmark for the Tiiny that shows what your box does under real work: prefill against prompt length, throughput over a long generation, what happens when several callers arrive at once, and what a reasoning model costs in wall time for the tokens nobody reads. Results are plain JSON and the report is one HTML file you can open anywhere. By default it benchmarks whatever is already loaded and touches nothing else.
farm install tiiny-bench, then farm start tiiny-bench and open http://localhost:8425
If you run it, post your numbers here. I'd like to see how the boxes compare.
I’m gonna try this soon and let you know