← CanIRunLocal · All hardware · All models
Yes, with partial offload (slower). Flint 40B-A2.5B (MoE) on RTX 5080 16GB at NVFP4 uses system RAM (24.6 GB needed, 15.5 GB usable), about 33 tok/s (estimate).
Hardware: RTX 5080 16GB · Model: Flint 40B-A2.5B (MoE)
RTX 5080 16GB has 16 GB of VRAM and 960 GB/s of memory bandwidth. The checker holds back 0.5 GB of VRAM for the driver and assumes 16 GB of system RAM, so fast memory usable by Ollama is 15.5 GB. Partial offload may also use 60% of that system RAM. Flint 40B-A2.5B (MoE) (39.8B, 2.5B active, Recursive) at NVFP4 is a 23 GB download and needs 24.6 GB with headroom for about 8k tokens.
| Best quant | NVFP4 |
|---|---|
| Download size | 23 GB |
| Memory needed | 24.6 GB |
| Usable fast memory | 15.5 GB |
| Fit | partial offload (slower) |
| Estimated speed | ~33 tok/s (estimate) |
| Command | ollama run flint:40b |
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