← CanIRunLocal · All hardware · All models
Yes. Bonsai 2 4B (ternary) fits in fast memory on Mac Studio M3 Ultra (60c GPU) 96GB at Q2_0 (2.8 GB needed, 72.0 GB usable), about 410 tok/s (estimate).
Hardware: Mac Studio M3 Ultra (60c GPU) 96GB · Model: Bonsai 2 4B (ternary)
Mac Studio M3 Ultra (60c GPU) 96GB has 96 GB of unified memory and 819 GB/s of memory bandwidth. Mac Studio M3 Ultra (80c GPU) 96GB has the same memory and bandwidth, so the fit numbers match; this page is Mac Studio M3 Ultra (60c GPU) 96GB. The checker leaves macOS a share of unified memory, so Ollama can use 72.0 GB. Bonsai 2 4B (ternary) (4B, Prism ML) at Q2_0 is a 1.6 GB download and needs 2.8 GB with headroom for about 8k tokens.
| Best quant | Q2_0 |
|---|---|
| Download size | 1.6 GB |
| Memory needed | 2.8 GB |
| Usable fast memory | 72.0 GB |
| Fit | fits in fast memory |
| Estimated speed | ~410 tok/s (estimate) |
| Command | ollama run bonsai-2:4b |
Estimates, not benchmarks. Model sizes are ollama.com download sizes; memory needed adds headroom for ~8k tokens of context; speeds are rough figures from memory bandwidth. Data updated 2026-10-23.
A project by OnDevice Labs · llms.txt