Models › LFM2.5 24B-A2B (MoE)
Can I run LFM2.5 24B-A2B (MoE) in Australia?
Yes. The cheapest hardware we track that holds LFM2.5 24B-A2B (MoE) at Q4_K_M with an 8k context is the AMD Radeon AI PRO R9700 32GB at A$2,299 (Mwave, checked 20 Oct 2026, 7:38 am AEST). Expect roughly 455 tokens/sec (estimate).
Memory needed by quantisation (8k context)
| Quant | Weights | Total needed | Cheapest hardware that fits |
|---|---|---|---|
Q4_K_M | 14.5 GB | 15.7 GB | AMD Radeon AI PRO R9700 32GB · A$2,299 |
Q5_K_M | 17.0 GB | 18.2 GB | AMD Radeon AI PRO R9700 32GB · A$2,299 |
Q6_K | 19.7 GB | 20.8 GB | AMD Radeon AI PRO R9700 32GB · A$2,299 |
Q8_0 | 25.5 GB | 26.6 GB | AMD Radeon AI PRO R9700 32GB · A$2,299 |
F16 | 47.9 GB | 49.1 GB |
All hardware that fits at Q4_K_M
| Hardware | Needs / usable | Est. tok/s | Price | Retailer link |
|---|---|---|---|---|
15.7 / 31.5 GB fits | 455 estimate | A$2,299 Mwave checked 20 Oct 2026, 7:38 am AEST | ||
15.7 / 18.0 GB fits | 211 estimate | A$2,699 Apple AU checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 18.0 GB fits | 207 estimate | A$2,999 Apple AU checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 23.5 GB fits | 535 estimate | A$3,599 Scorptec · not in stock online checked 20 Oct 2026, 7:41 am AEST | ||
15.7 / 48.0 GB fits | 156 estimate | A$3,915 Minisforum AU · not in stock online checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 27.0 GB fits | 281 estimate | A$4,299 Apple AU checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 96.0 GB fits | 156 estimate | A$5,719 Minisforum AU checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 48.0 GB fits | 156 estimate | A$5,894 Mwave checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 48.0 GB fits | 184 estimate | A$7,149 Microsoft Store AU · stale checked 19 Oct 2026, 8:18 pm AEST | ||
15.7 / 96.0 GB fits | 156 estimate | A$8,999 Mwave · not in stock online checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 96.0 GB fits | 179 estimate | A$8,999 Mwave checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 72.0 GB fits | 733 estimate | A$9,499 Apple AU checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 96.0 GB fits | 184 estimate | A$9,649 Microsoft Store AU · stale checked 19 Oct 2026, 8:18 pm AEST | ||
15.7 / 31.5 GB fits | 907 estimate | A$10,999 Scorptec checked 20 Oct 2026, 7:41 am AEST | ||
15.7 / 192.0 GB fits | 560 estimate | A$11,449 Apple AU checked 20 Oct 2026, 7:44 am AEST | ||
15.7 / 95.5 GB fits | 857 estimate | A$27,599 Scorptec checked 20 Oct 2026, 7:42 am AEST | ||
15.7 / 23.5 GB fits | 495 estimate | price unavailable | ||
15.7 / 23.5 GB fits | 582 estimate | price unavailable | ||
15.7 / 23.5 GB fits | 571 estimate | price unavailable | ||
15.7 / 36.0 GB fits | 375 estimate | price unavailable | ||
15.7 / 48.0 GB fits | 156 estimate | price unavailable | ||
15.7 / 96.0 GB fits | 156 estimate | price unavailable |
The tok/s figures are bandwidth-based estimates for this model. Where a device has a measured benchmark, the estimate is calibrated to that run. They are not measurements. Methodology.
Graphics card + system RAM (MoE experts in RAM)
Mixture-of-experts models only read a few experts per token, so llama.cpp can keep the expert weights in system RAM and everything else on the GPU (e.g. --n-cpu-moe / -ot exps=CPU). That lets a 12–32GB card run a much bigger model.
All figures below are estimates (Q4_K_M, 8k context, dual-channel DDR5-5600). How we estimate.
| Graphics card | Card price | GPU only | + 32GB RAM | + 64GB RAM | + 128GB RAM |
|---|---|---|---|---|---|
A$419 | no | fits 5 GB in RAM · ~105 tok/s | fits 5 GB in RAM · ~105 tok/s | fits 5 GB in RAM · ~105 tok/s | |
A$539 | no | fits 5 GB in RAM · ~108 tok/s | fits 5 GB in RAM · ~108 tok/s | fits 5 GB in RAM · ~108 tok/s | |
A$699 | no | fits 2 GB in RAM · ~158 tok/s | fits 2 GB in RAM · ~158 tok/s | fits 2 GB in RAM · ~158 tok/s | |
A$1,099 | no | fits 2 GB in RAM · ~191 tok/s | fits 2 GB in RAM · ~191 tok/s | fits 2 GB in RAM · ~191 tok/s | |
A$1,099 | no | fits 2 GB in RAM · ~224 tok/s | fits 2 GB in RAM · ~224 tok/s | fits 2 GB in RAM · ~224 tok/s | |
A$1,169 | no | fits 2 GB in RAM · ~241 tok/s | fits 2 GB in RAM · ~241 tok/s | fits 2 GB in RAM · ~241 tok/s | |
A$1,249 | no | fits 5 GB in RAM · ~125 tok/s | fits 5 GB in RAM · ~125 tok/s | fits 5 GB in RAM · ~125 tok/s | |
A$1,699 | no | fits 2 GB in RAM · ~272 tok/s | fits 2 GB in RAM · ~272 tok/s | fits 2 GB in RAM · ~272 tok/s | |
A$2,299 | no | fits 2 GB in RAM · ~276 tok/s | fits 2 GB in RAM · ~276 tok/s | fits 2 GB in RAM · ~276 tok/s | |
A$2,299 | fits ~455 tok/s | fits in VRAM | fits in VRAM | fits in VRAM | |
A$3,599 | fits ~535 tok/s | fits in VRAM | fits in VRAM | fits in VRAM | |
A$10,999 | fits ~907 tok/s | fits in VRAM | fits in VRAM | fits in VRAM | |
A$27,599 | fits ~857 tok/s | fits in VRAM | fits in VRAM | fits in VRAM | |
n/a | no | fits 2 GB in RAM · ~139 tok/s | fits 2 GB in RAM · ~139 tok/s | fits 2 GB in RAM · ~139 tok/s | |
n/a | fits ~495 tok/s | fits in VRAM | fits in VRAM | fits in VRAM | |
n/a | fits ~582 tok/s | fits in VRAM | fits in VRAM | fits in VRAM | |
n/a | fits ~571 tok/s | fits in VRAM | fits in VRAM | fits in VRAM |
DDR5 kit prices today: 32GB: A$610 (Mwave, checked 20 Oct 2026, 7:44 am AEST) · 64GB: A$1,499 (Mwave, checked 20 Oct 2026, 7:44 am AEST) · 128GB: A$2,511 (Mwave, not in stock online, checked 20 Oct 2026, 7:44 am AEST). Card + kit only; the rest of the PC isn't included. RAM-offload speeds are estimates and can be optimistic, so check community results below.
Community measured
No reader results yet. Be the first to submit one →
Try other quantisations and context lengths in the Can I run it? tool → · Submit your own LFM2.5 24B-A2B (MoE) benchmark →