← CanIRunLocal · All hardware · All models
Yes. Flint 40B-A2.5B (MoE) fits in fast memory on M3 Max (40c GPU) 64GB at Q4_1 (27.6 GB needed, 48.0 GB usable), about 98 tok/s (estimate).
Hardware: M3 Max (40c GPU) 64GB · Model: Flint 40B-A2.5B (MoE)
M3 Max (40c GPU) 64GB has 64 GB of unified memory and 400 GB/s of memory bandwidth. M1 Max (32c GPU) 64GB and M2 Max (30c GPU) 64GB and M2 Max (38c GPU) 64GB have the same memory and bandwidth, so the fit numbers match; this page is M3 Max (40c GPU) 64GB. The checker leaves macOS a share of unified memory, so Ollama can use 48.0 GB. Flint 40B-A2.5B (MoE) (39.8B, 2.5B active, Recursive) at Q4_1 is a 26 GB download and needs 27.6 GB with headroom for about 8k tokens.
| Best quant | Q4_1 |
|---|---|
| Download size | 26 GB |
| Memory needed | 27.6 GB |
| Usable fast memory | 48.0 GB |
| Fit | fits in fast memory |
| Estimated speed | ~98 tok/s (estimate) |
| Command | ollama run flint:40b-a2b-q4_1 |
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