RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run LFM2.5 VL 3B locally?

LFM2.5 VL 3B by LiquidAI has 3.1B parameters. At 4-bit with an 8K context it needs about 2.3 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track. See all models that fit in 8 GB of VRAM.

2.3 GBQ4_K_M · 8K context
RTX 4060smallest consumer GPU that fits
M1 16GBsmallest Mac that fits
32Kmax context
GQAattention type

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from LiquidAI/LFM2.5-VL-3B-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx
2-bit (Q2_K)rough1.8 GB1.9 GB2.2 GB
3-bit (Q3_K_M)usable2.1 GB2.2 GB2.6 GB
4-bit (Q4_K_M) •good, most popular2.3 GB2.3 GB2.7 GB
5-bit (Q5_K_M) •very good2.5 GB2.6 GB3.0 GB
6-bit (Q6_K) •near lossless2.8 GB2.8 GB3.2 GB
8-bit (Q8_0) •effectively lossless3.4 GB3.4 GB3.8 GB
16-bit (F16/BF16) •original weights10.8 GB10.8 GB11.2 GB

Which GPUs and Macs run LFM2.5 VL 3B?

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 GBFitsQ8_03.4 GB125details
RTX 4060 Ti 16GB16 GBFitsF1610.8 GB27details
RTX 309024 GBFitsF1610.8 GB87details
RTX 409024 GBFitsF1610.8 GB93details
RTX 509032 GBFitsF1610.8 GB166details
RTX 5070 Ti16 GBFitsF1610.8 GB83details
RX 7900 XTX24 GBFitsF1610.8 GB89details
Mac M4 Pro 48GB48 GBFitsF1610.8 GB25details
Mac M4 Max 128GB128 GBFitsF1610.8 GB51details
Mac M3 Ultra 512GB512 GBFitsF1610.8 GB76details
DGX Spark128 GBFitsF1610.8 GB25details
RTX 407012 GBFitsQ8_03.4 GB175details
RTX 5060 Ti 16GB16 GBFitsF1610.8 GB41details
RTX 40608 GBFitsQ8_03.4 GB95details
Arc B58012 GBFitsQ8_03.4 GB159details
RTX 3060 Ti8 GBFitsQ8_03.4 GB156details
RTX 30708 GBFitsQ8_03.4 GB156details
RTX 3080 10GB10 GBFitsQ8_03.4 GB264details
RTX 3080 12GB12 GBFitsQ8_03.4 GB317details
RTX 3090 Ti24 GBFitsF1610.8 GB93details
RTX 4060 Ti 8GB8 GBFitsQ8_03.4 GB100details
RTX 4070 Super12 GBFitsQ8_03.4 GB175details
RTX 4070 Ti12 GBFitsQ8_03.4 GB175details
RTX 4070 Ti Super16 GBFitsF1610.8 GB62details
RTX 408016 GBFitsF1610.8 GB66details
RTX 4080 Super16 GBFitsF1610.8 GB68details
RTX 50608 GBFitsQ8_03.4 GB156details
RTX 507012 GBFitsQ8_03.4 GB234details
RTX 508016 GBFitsF1610.8 GB89details
2x RTX 30902×24 GBFitsF1610.8 GB87details
2x RTX 40902×24 GBFitsF1610.8 GB93details
2x RTX 50902×32 GBFitsF1610.8 GB166details
RTX A600048 GBFitsF1610.8 GB71details
RTX 6000 Ada48 GBFitsF1610.8 GB89details
RTX PRO 6000 Blackwell96 GBFitsF1610.8 GB166details
L40S48 GBFitsF1610.8 GB80details
A100 40GB40 GBFitsF1610.8 GB144details
A100 80GB80 GBFitsF1610.8 GB189details
H100 PCIe80 GBFitsF1610.8 GB185details
H100 SXM80 GBFitsF1610.8 GB310details
H200141 GBFitsF1610.8 GB444details
RX 76008 GBFitsQ8_03.4 GB100details
RX 7700 XT12 GBFitsQ8_03.4 GB150details
RX 7800 XT16 GBFitsF1610.8 GB58details
RX 7900 GRE16 GBFitsF1610.8 GB53details
RX 7900 XT20 GBFitsF1610.8 GB74details
RX 907016 GBFitsF1610.8 GB59details
RX 9070 XT16 GBFitsF1610.8 GB59details
Ryzen AI Max+ 395 128GB128 GBFitsF1610.8 GB24details
Arc A770 16GB16 GBFitsF1610.8 GB52details
Mac M1 16GB16 GBFitsQ8_03.4 GB24details
Mac M2 24GB24 GBFitsF1610.8 GB9.3details
Mac M3 24GB24 GBFitsF1610.8 GB9.3details
Mac M4 16GB16 GBFitsQ8_03.4 GB42details
Mac M4 32GB32 GBFitsF1610.8 GB11details
Mac M1 Pro 32GB32 GBFitsF1610.8 GB19details
Mac M3 Pro 36GB36 GBFitsF1610.8 GB14details
Mac M4 Pro 24GB24 GBFitsF1610.8 GB25details
Mac M4 Pro 64GB64 GBFitsF1610.8 GB25details
Mac M1 Max 64GB64 GBFitsF1610.8 GB37details
Mac M2 Max 96GB96 GBFitsF1610.8 GB37details
Mac M3 Max 128GB128 GBFitsF1610.8 GB37details
Mac M4 Max 64GB64 GBFitsF1610.8 GB51details
Mac M2 Ultra 192GB192 GBFitsF1610.8 GB74details
Mac M3 Ultra 256GB256 GBFitsF1610.8 GB76details

Questions

How much VRAM does LFM2.5 VL 3B need?

About 2.3 GB at 4-bit (Q4_K_M) with an 8K context: 1.6 GB of weights (real GGUF file size), 0.1 GB of KV cache and 0.6 GB of runtime overhead. At 8-bit it needs about 3.4 GB.

What is the smallest GPU that runs LFM2.5 VL 3B?

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 LFM2.5 VL 3B 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.

Similar models

Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.

d1 3B

3.1B · 2.3 GB at 4-bit

LFM2.5 2.6B

2.7B · 2.3 GB at 4-bit

d1 2B

1.9B · 1.8 GB at 4-bit

Qwen2.5 3B Instruct

3.1B · 2.9 GB at 4-bit

SmolLM3 3B

3.1B · 3.0 GB at 4-bit

SmolLM3 3B Base

3.1B · 3.0 GB at 4-bit

Model facts

Hugging Face repoLiquidAI/LFM2.5-VL-3B
Parameters3,123,483,888 (3.1B)
Layers / hidden size30 / 2048
AttentionGQA (hybrid: 22 linear/state layers); 32 heads, 8 KV heads, head dim 64
KV cache per token (FP16)16.0 KiB
Max context32,768 tokens
Licenseother
Released on HFAug 11, 2026
Downloads (30d) / likes17,704 / 238

Sources: Hugging Face model API · config.json (LiquidAI/LFM2.5-VL-3B) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.