Models › LFM2.5-24B-A2B

LFM2.5-24B-A2B: VRAM needs and speed by GPU

At UD-Q4_K_XL, LFM2.5-24B-A2B needs about 16.3 GB of VRAM. No measured runs yet. Speeds below are estimates. Submit a run.

23.97B parameters (mixture-of-experts, ~2.0B active) · license: other · 131,072-token context · GGUF published Oct 13, 2026 · GGUF files on Hugging Face

Memory needed per quantization

QuantFile sizeNeeds (estimate)Smallest card it fits
UD-IQ2_XXS7.5 GB8.9 GB12 GB cards
UD-IQ2_M7.5 GB8.9 GB12 GB cards
UD-Q2_K_XL8.0 GB9.4 GB12 GB cards
UD-IQ3_XXS9.0 GB10.4 GB12 GB cards
UD-Q3_K_XL10.9 GB12.3 GB16 GB cards
UD-Q4_K_XL14.9 GB16.3 GB24 GB cards
UD-Q5_K_XL17.9 GB19.3 GB24 GB cards
UD-Q6_K_XL21.9 GB23.3 GB24 GB cards
Q8_025.5 GB26.9 GB32 GB cards
BF1648.0 GB49.4 GB96 GB cards

Expected speed on popular hardware

HardwareBest quant that fitsExpected tok/s
NVIDIA RTX 3060UD-IQ3_XXS206 estimate
NVIDIA RTX 4060 TiDoes not fit
NVIDIA RTX 3090UD-IQ3_XXS439 estimate
NVIDIA RTX 4090UD-IQ3_XXS487 estimate
NVIDIA RTX 5090Q8_0270 estimate
AMD Radeon RX 7900 XTXUD-IQ3_XXS489 estimate
Apple M4 Pro (20-core GPU) (64 GB)Q8_054.5 estimate
Apple M3 Max (40-core GPU) (64 GB)Q8_079.6 estimate
AMD Ryzen AI Max+ 395 (Radeon 8060S)Q8_048.9 estimate

Check your own card

Too big for your card?

Run LFM2.5-24B-A2B on a rented 24-80 GB GPU by the hour.

Rent a bigger GPU on RunPod

Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.