RunsOnMy ModelsGPUs & MacsAPI pricesVRAM calculatorNew

Can I run LFM2 350M locally?

LFM2 350M by LiquidAI has 354M parameters. At 4-bit with an 8K context it needs about 1.0 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track.

1.0 GBQ4_K_M · 8K context
RTX 4060smallest consumer GPU that fits
M1 16GBsmallest Mac that fits
125Kmax context
GQAattention type

GGUF quant file sizes were not in size order, so this page uses size estimates instead of real file sizes.

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx
2-bit (Q2_K)rough0.8 GB0.9 GB1.6 GB
3-bit (Q3_K_M)usable0.8 GB0.9 GB1.7 GB
4-bit (Q4_K_M)good, most popular0.9 GB1.0 GB1.7 GB
5-bit (Q5_K_M)very good0.9 GB1.0 GB1.8 GB
6-bit (Q6_K)near lossless0.9 GB1.0 GB1.8 GB
8-bit (Q8_0)effectively lossless1.0 GB1.1 GB1.9 GB
16-bit (F16/BF16)original weights1.3 GB1.4 GB2.2 GB

Which GPUs and Macs run LFM2 350M?

Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).

HardwareMemoryFits?Best quantNeedsMax tok/s (est.)
RTX 3060 12GB12 GBFitsF161.4 GB508details
RTX 4060 Ti 16GB16 GBFitsF161.4 GB406details
RTX 309024 GBFitsF161.4 GB1320details
RTX 409024 GBFitsF161.4 GB1422details
RTX 509032 GBFitsF161.4 GB2528details
RTX 5070 Ti16 GBFitsF161.4 GB1264details
RX 7900 XTX24 GBFitsF161.4 GB1354details
Mac M4 Pro 48GB48 GBFitsF161.4 GB385details
Mac M4 Max 128GB128 GBFitsF161.4 GB770details
Mac M3 Ultra 512GB512 GBFitsF161.4 GB1155details
DGX Spark128 GBFitsF161.4 GB385details
RTX 407012 GBFitsF161.4 GB711details
RTX 5060 Ti 16GB16 GBFitsF161.4 GB632details
RTX 40608 GBFitsF161.4 GB384details
Arc B58012 GBFitsF161.4 GB643details
RTX 3060 Ti8 GBFitsF161.4 GB632details
RTX 30708 GBFitsF161.4 GB632details
RTX 3080 10GB10 GBFitsF161.4 GB1072details
RTX 3080 12GB12 GBFitsF161.4 GB1286details
RTX 3090 Ti24 GBFitsF161.4 GB1422details
RTX 4060 Ti 8GB8 GBFitsF161.4 GB406details
RTX 4070 Super12 GBFitsF161.4 GB711details
RTX 4070 Ti12 GBFitsF161.4 GB711details
RTX 4070 Ti Super16 GBFitsF161.4 GB948details
RTX 408016 GBFitsF161.4 GB1011details
RTX 4080 Super16 GBFitsF161.4 GB1038details
RTX 50608 GBFitsF161.4 GB632details
RTX 507012 GBFitsF161.4 GB948details
RTX 508016 GBFitsF161.4 GB1354details
2x RTX 30902×24 GBFitsF161.4 GB1320details
2x RTX 40902×24 GBFitsF161.4 GB1422details
2x RTX 50902×32 GBFitsF161.4 GB2528details
RTX A600048 GBFitsF161.4 GB1083details
RTX 6000 Ada48 GBFitsF161.4 GB1354details
RTX PRO 6000 Blackwell96 GBFitsF161.4 GB2528details
L40S48 GBFitsF161.4 GB1219details
A100 40GB40 GBFitsF161.4 GB2193details
A100 80GB80 GBFitsF161.4 GB2876details
H100 PCIe80 GBFitsF161.4 GB2821details
H100 SXM80 GBFitsF161.4 GB4725details
H200141 GBFitsF161.4 GB6770details
RX 76008 GBFitsF161.4 GB406details
RX 7700 XT12 GBFitsF161.4 GB609details
RX 7800 XT16 GBFitsF161.4 GB880details
RX 7900 GRE16 GBFitsF161.4 GB812details
RX 7900 XT20 GBFitsF161.4 GB1128details
RX 907016 GBFitsF161.4 GB903details
RX 9070 XT16 GBFitsF161.4 GB903details
Ryzen AI Max+ 395 128GB128 GBFitsF161.4 GB361details
Arc A770 16GB16 GBFitsF161.4 GB790details
Mac M1 16GB16 GBFitsF161.4 GB96details
Mac M2 24GB24 GBFitsF161.4 GB141details
Mac M3 24GB24 GBFitsF161.4 GB141details
Mac M4 16GB16 GBFitsF161.4 GB169details
Mac M4 32GB32 GBFitsF161.4 GB169details
Mac M1 Pro 32GB32 GBFitsF161.4 GB282details
Mac M3 Pro 36GB36 GBFitsF161.4 GB212details
Mac M4 Pro 24GB24 GBFitsF161.4 GB385details
Mac M4 Pro 64GB64 GBFitsF161.4 GB385details
Mac M1 Max 64GB64 GBFitsF161.4 GB564details
Mac M2 Max 96GB96 GBFitsF161.4 GB564details
Mac M3 Max 128GB128 GBFitsF161.4 GB564details
Mac M4 Max 64GB64 GBFitsF161.4 GB770details
Mac M2 Ultra 192GB192 GBFitsF161.4 GB1128details
Mac M3 Ultra 256GB256 GBFitsF161.4 GB1155details

Questions

How much VRAM does LFM2 350M need?

About 1.0 GB at 4-bit (Q4_K_M) with an 8K context: 0.2 GB of weights (estimated), 0.3 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 1.1 GB.

What is the smallest GPU that runs LFM2 350M?

At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 4060 (8 GB).

Can I run LFM2 350M on a Mac?

Yes. At Q4_K_M with 8K context it fits on a Mac M1 16GB or larger, using the default macOS GPU memory limit.

Model facts

Hugging Face repoLiquidAI/LFM2-350M
Parameters354,483,968 (354M)
Layers / hidden size16 / 1024
AttentionGQA; 16 heads, 8 KV heads, head dim 64
KV cache per token (FP16)32.0 KiB
Max context128,000 tokens
Licenseother
Released on HFJul 10, 2025
Downloads (30d) / likes15,966 / 260

Sources: Hugging Face model API · config.json (LiquidAI/LFM2-350M) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.