How much VRAM does LFM2.5 2.6B need?

At Q4_K_M the weights are 1.5 GB; with an 8K context KV cache and runtime overhead it needs 2.2 GB, so a 8 GB card runs it with headroom.

Can a GeForce RTX 4070 run LFM2.5 2.6B?

Yes — Q8_0 fits with headroom at 8K context, about 114 tokens per second estimated.

Can a GeForce RTX 4090 run LFM2.5 2.6B?

Yes — Q8_0 fits with headroom at 8K context, about 228 tokens per second estimated.

Can a GeForce RTX 3060 12 GB run LFM2.5 2.6B?

Yes — Q8_0 fits with headroom at 8K context, about 82 tokens per second estimated.

Can a M3 Max · 64 GB run LFM2.5 2.6B?

Yes — Q8_0 fits with headroom at 8K context, about 84 tokens per second estimated.

Hardware requirements

What hardware do I need to run LFM2.5 2.6B?

2.7B 128K context LFM Open License v1.0 New this month

At Q4_K_M the weights are 1.5 GB; with an 8K KV cache and runtime overhead that is 2.2 GB, so it wants a 8 GB card to run with headroom — in practice 8 GB VRAM: GeForce RTX 3070 Ti and 7 others. Every figure below is computed from the model’s published shape against each device’s usable memory; tokens per second are estimates. The method.

Minimum practical 8 GB VRAM

GeForce RTX 3070 Ti — Q8_0, 3.4 GB of 7.0 GB, ~138 tok/s, up to 128K context.

Also: Arc A750, GeForce RTX 5060, GeForce RTX 3070, GeForce RTX 3060 Ti, GeForce RTX 4060 Ti +2 more

the smallest card that holds it at all, at Q3_K_M — the lowest quantisation worth running

Recommended 8 GB VRAM

GeForce RTX 3070 Ti — Q4_K_M, 2.2 GB of 7.0 GB, ~242 tok/s, up to 128K context.

Also: Arc A750, GeForce RTX 5060, GeForce RTX 3070, GeForce RTX 3060 Ti, GeForce RTX 4060 Ti +2 more

runs the recommended quantisation with headroom at 8K context

High quality 8 GB VRAM

GeForce RTX 3070 Ti — Q8_0, 3.4 GB of 7.0 GB, ~138 tok/s, up to 128K context.

Also: Arc A750, GeForce RTX 5060, GeForce RTX 3070, GeForce RTX 3060 Ti, GeForce RTX 4060 Ti +2 more

near-lossless Q8_0 with headroom

Long context 8 GB VRAM

GeForce RTX 3070 Ti — Q4_K_M, 4.1 GB of 7.0 GB, ~242 tok/s at 128K.

Also: Arc A750, GeForce RTX 5060, GeForce RTX 3070, GeForce RTX 3060 Ti, GeForce RTX 4060 Ti +2 more

128K tokens (or the model’s maximum) at the recommended quantisation

Apple Silicon 16 GB unified

M1 · 16 GB — Q4_K_M, 2.2 GB of 10.7 GB, ~25 tok/s, up to 128K context.

smallest Mac that runs the recommended quantisation with headroom

CPU only 32 GB RAM

CPU only · DDR4 dual-channel — Q8_0, 3.4 GB of 25.6 GB, ~8.0 tok/s, up to 128K context.

no GPU — weights stream from system RAM

How much VRAM at each quantisation

Quant	Weights	+KV 8K	+KV 32K	+KV 128K	Total @ 8K	Needs	Quality
F16	5.0 GB	0.13 GB	0.50 GB	2.0 GB	5.8 GB	8 GB card	Reference
Q8_0	2.7 GB	0.13 GB	0.50 GB	2.0 GB	3.4 GB	8 GB card	−0.1% ppl
Q6_K	2.1 GB	0.13 GB	0.50 GB	2.0 GB	2.8 GB	8 GB card	−0.4% ppl
Q5_K_M	1.8 GB	0.13 GB	0.50 GB	2.0 GB	2.5 GB	8 GB card	−0.8% ppl
Q4_K_M	1.5 GB	0.13 GB	0.50 GB	2.0 GB	2.2 GB	8 GB card	−1.9% ppl
Q3_K_M	1.2 GB	0.13 GB	0.50 GB	2.0 GB	2.0 GB	8 GB card	−5.4% ppl
Q2_K	1.1 GB	0.13 GB	0.50 GB	2.0 GB	1.8 GB	8 GB card	−15% ppl

Totals add 0.6 GB runtime overhead and an f16 KV cache; "needs" is the smallest discrete card class whose usable memory (after display and driver reserve) leaves 15% headroom. A q8_0 cache roughly halves the KV columns.

Every device, checked

System RAM for offload:

16 32 64 128 256

NVIDIA · GeForce

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
GeForce RTX 5090	32 GB	Q8_0	Q8_0	128K	715	yes	Check
GeForce RTX 5080	16 GB	Q8_0	Q8_0	128K	383	yes	Check
GeForce RTX 5070 Ti	16 GB	Q8_0	Q8_0	128K	357	yes	Check
GeForce RTX 5070	12 GB	Q8_0	Q8_0	128K	268	yes	Check
GeForce RTX 5060 Ti 16 GB	16 GB	Q8_0	Q8_0	128K	179	yes	Check
GeForce RTX 5060	8 GB	Q8_0	Q8_0	128K	179	yes	Check
GeForce RTX 4090	24 GB	Q8_0	Q8_0	128K	402	yes	Check
GeForce RTX 4080 Super	16 GB	Q8_0	Q8_0	128K	293	yes	Check
GeForce RTX 4080	16 GB	Q8_0	Q8_0	128K	286	yes	Check
GeForce RTX 4070 Ti Super	16 GB	Q8_0	Q8_0	128K	268	yes	Check
GeForce RTX 4070 Ti	12 GB	Q8_0	Q8_0	128K	201	yes	Check
GeForce RTX 4070 Super	12 GB	Q8_0	Q8_0	128K	201	yes	Check
GeForce RTX 4070	12 GB	Q8_0	Q8_0	128K	201	yes	Check
GeForce RTX 4060 Ti 16 GB	16 GB	Q8_0	Q8_0	128K	115	yes	Check
GeForce RTX 4060 Ti	8 GB	Q8_0	Q8_0	128K	115	yes	Check
GeForce RTX 4060	8 GB	Q8_0	Q8_0	128K	108	yes	Check
GeForce RTX 3090 Ti	24 GB	Q8_0	Q8_0	128K	402	yes	Check
GeForce RTX 3090	24 GB	Q8_0	Q8_0	128K	373	yes	Check
GeForce RTX 3080 Ti	12 GB	Q8_0	Q8_0	128K	364	yes	Check
GeForce RTX 3080 12 GB	12 GB	Q8_0	Q8_0	128K	364	yes	Check
GeForce RTX 3080	10 GB	Q8_0	Q8_0	128K	303	yes	Check
GeForce RTX 3070 Ti	8 GB	Q8_0	Q8_0	128K	242	yes	Check
GeForce RTX 3070	8 GB	Q8_0	Q8_0	128K	179	yes	Check
GeForce RTX 3060 Ti	8 GB	Q8_0	Q8_0	128K	179	yes	Check
GeForce RTX 3060 12 GB	12 GB	Q8_0	Q8_0	128K	144	yes	Check
GeForce RTX 2080 Ti	11 GB	Q8_0	Q8_0	128K	246	yes	Check
GeForce RTX 2060 12 GB	12 GB	Q8_0	Q8_0	128K	134	yes	Check
GeForce GTX 1080 Ti	11 GB	Q8_0	Q8_0	128K	193	yes	Check

NVIDIA · GeForce laptop

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
GeForce RTX 5090 Laptop	24 GB	Q8_0	Q8_0	128K	357	yes	Check
GeForce RTX 4090 Laptop	16 GB	Q8_0	Q8_0	128K	230	yes	Check
GeForce RTX 4080 Laptop	12 GB	Q8_0	Q8_0	128K	172	yes	Check
GeForce RTX 4070 Laptop	8 GB	Q8_0	Q8_0	128K	102	yes	Check

NVIDIA · Workstation

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
RTX 6000 Ada	48 GB	Q8_0	Q8_0	128K	383	yes	Check
RTX 5000 Ada	32 GB	Q8_0	Q8_0	128K	230	yes	Check
RTX 4000 Ada	20 GB	Q8_0	Q8_0	128K	144	yes	Check
RTX 2000 Ada	16 GB	Q8_0	Q8_0	128K	89	yes	Check
RTX A6000	48 GB	Q8_0	Q8_0	128K	306	yes	Check
RTX A5000	24 GB	Q8_0	Q8_0	128K	306	yes	Check
RTX A4000	16 GB	Q8_0	Q8_0	128K	179	yes	Check

NVIDIA · Data centre

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
L40S	48 GB	Q8_0	Q8_0	128K	345	yes	Check
L4	24 GB	Q8_0	Q8_0	128K	120	yes	Check
A10	24 GB	Q8_0	Q8_0	128K	239	yes	Check
A100 40 GB	40 GB	Q8_0	Q8_0	128K	620	yes	Check
A100 80 GB	80 GB	Q8_0	Q8_0	128K	813	yes	Check
H100 PCIe	80 GB	Q8_0	Q8_0	128K	797	yes	Check
H100 SXM	80 GB	Q8_0	Q8_0	128K	1336	yes	Check
H200 SXM	141 GB	Q8_0	Q8_0	128K	1914	yes	Check
Tesla P40	24 GB	Q8_0	Q8_0	128K	138	yes	Check

AMD · Radeon

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Radeon RX 9070 XT	16 GB	Q8_0	Q8_0	128K	257	yes	Check
Radeon RX 9070	16 GB	Q8_0	Q8_0	128K	257	yes	Check
Radeon RX 7900 XTX	24 GB	Q8_0	Q8_0	128K	383	yes	Check
Radeon RX 7900 XT	20 GB	Q8_0	Q8_0	128K	319	yes	Check
Radeon RX 7900 GRE	16 GB	Q8_0	Q8_0	128K	230	yes	Check
Radeon RX 7800 XT	16 GB	Q8_0	Q8_0	128K	249	yes	Check
Radeon RX 7700 XT	12 GB	Q8_0	Q8_0	128K	172	yes	Check
Radeon RX 7600 XT	16 GB	Q8_0	Q8_0	128K	115	yes	Check
Radeon RX 6900 XT	16 GB	Q8_0	Q8_0	128K	204	yes	Check
Radeon RX 6800 XT	16 GB	Q8_0	Q8_0	128K	204	yes	Check
Radeon RX 6700 XT	12 GB	Q8_0	Q8_0	128K	153	yes	Check

AMD · Radeon Pro

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Radeon Pro W7900	48 GB	Q8_0	Q8_0	128K	345	yes	Check
Radeon Pro W7800	32 GB	Q8_0	Q8_0	128K	230	yes	Check

AMD · Instinct

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Instinct MI210	64 GB	Q8_0	Q8_0	128K	653	yes	Check
Instinct MI300X	192 GB	Q8_0	Q8_0	128K	2113	yes	Check

AMD · Ryzen AI Max

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Ryzen AI Max+ 395 · 128 GB	128 GB	Q8_0	Q8_0	128K	94	yes	Check
Ryzen AI Max+ 395 · 64 GB	64 GB	Q8_0	Q8_0	128K	94	yes	Check

Apple · Apple Silicon

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
M1 · 16 GB	16 GB	Q8_0	Q8_0	128K	25	yes	Check
M1 Pro · 32 GB	32 GB	Q8_0	Q8_0	128K	74	yes	Check
M1 Max · 64 GB	64 GB	Q8_0	Q8_0	128K	147	yes	Check
M1 Ultra · 128 GB	128 GB	Q8_0	Q8_0	128K	294	yes	Check
M2 · 24 GB	24 GB	Q8_0	Q8_0	128K	37	yes	Check
M2 Pro · 32 GB	32 GB	Q8_0	Q8_0	128K	74	yes	Check
M2 Max · 96 GB	96 GB	Q8_0	Q8_0	128K	147	yes	Check
M2 Ultra · 192 GB	192 GB	Q8_0	Q8_0	128K	294	yes	Check
M3 · 24 GB	24 GB	Q8_0	Q8_0	128K	37	yes	Check
M3 Pro · 36 GB	36 GB	Q8_0	Q8_0	128K	55	yes	Check
M3 Max · 64 GB	64 GB	Q8_0	Q8_0	128K	147	yes	Check
M3 Max · 128 GB	128 GB	Q8_0	Q8_0	128K	147	yes	Check
M3 Ultra · 256 GB	256 GB	Q8_0	Q8_0	128K	294	yes	Check
M3 Ultra · 512 GB	512 GB	Q8_0	Q8_0	128K	294	yes	Check
M4 · 32 GB	32 GB	Q8_0	Q8_0	128K	44	yes	Check
M4 Pro · 48 GB	48 GB	Q8_0	Q8_0	128K	100	yes	Check
M4 Pro · 64 GB	64 GB	Q8_0	Q8_0	128K	100	yes	Check
M4 Max · 64 GB	64 GB	Q8_0	Q8_0	128K	151	yes	Check
M4 Max · 128 GB	128 GB	Q8_0	Q8_0	128K	201	yes	Check

Intel · Arc

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Arc B580	12 GB	Q8_0	Q8_0	128K	182	yes	Check
Arc B570	10 GB	Q8_0	Q8_0	128K	152	yes	Check
Arc A770 16 GB	16 GB	Q8_0	Q8_0	128K	223	yes	Check
Arc A750	8 GB	Q8_0	Q8_0	128K	204	yes	Check

CPU · CPU + system RAM

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
CPU only · DDR4 dual-channel	32 GB	Q8_0	Q8_0	128K	14	yes	Check
CPU only · DDR5 dual-channel	64 GB	Q8_0	Q8_0	128K	25	yes	Check
CPU only · DDR5 8-channel server	256 GB	Q8_0	Q8_0	128K	85	yes	Check