RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run SmolVLM2 500M Video Instruct locally?

SmolVLM2 500M Video Instruct by HuggingFaceTB has 507M parameters. At 4-bit with an 8K context it needs about 1.1 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.

1.1 GBQ4_K_M · 8K context
RTX 4060smallest consumer GPU that fits
M1 16GBsmallest Mac that fits
8Kmax 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 ggml-org/SmolVLM2-500M-Video-Instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx
2-bit (Q2_K)rough0.8 GB1.0 GB
3-bit (Q3_K_M)usable0.9 GB1.0 GB
4-bit (Q4_K_M)good, most popular0.9 GB1.1 GB
5-bit (Q5_K_M)very good1.0 GB1.1 GB
6-bit (Q6_K)near lossless1.0 GB1.2 GB
8-bit (Q8_0) •effectively lossless1.1 GB1.2 GB
16-bit (F16/BF16) •original weights1.4 GB1.6 GB

Which GPUs and Macs run SmolVLM2 500M Video Instruct?

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 GBFitsF161.6 GB439details
RTX 4060 Ti 16GB16 GBFitsF161.6 GB351details
RTX 309024 GBFitsF161.6 GB1141details
RTX 409024 GBFitsF161.6 GB1229details
RTX 509032 GBFitsF161.6 GB2184details
RTX 5070 Ti16 GBFitsF161.6 GB1092details
RX 7900 XTX24 GBFitsF161.6 GB1170details
Mac M4 Pro 48GB48 GBFitsF161.6 GB333details
Mac M4 Max 128GB128 GBFitsF161.6 GB666details
Mac M3 Ultra 512GB512 GBFitsF161.6 GB998details
DGX Spark128 GBFitsF161.6 GB333details
RTX 407012 GBFitsF161.6 GB614details
RTX 5060 Ti 16GB16 GBFitsF161.6 GB546details
RTX 40608 GBFitsF161.6 GB332details
Arc B58012 GBFitsF161.6 GB556details
RTX 3060 Ti8 GBFitsF161.6 GB546details
RTX 30708 GBFitsF161.6 GB546details
RTX 3080 10GB10 GBFitsF161.6 GB926details
RTX 3080 12GB12 GBFitsF161.6 GB1112details
RTX 3090 Ti24 GBFitsF161.6 GB1229details
RTX 4060 Ti 8GB8 GBFitsF161.6 GB351details
RTX 4070 Super12 GBFitsF161.6 GB614details
RTX 4070 Ti12 GBFitsF161.6 GB614details
RTX 4070 Ti Super16 GBFitsF161.6 GB819details
RTX 408016 GBFitsF161.6 GB874details
RTX 4080 Super16 GBFitsF161.6 GB897details
RTX 50608 GBFitsF161.6 GB546details
RTX 507012 GBFitsF161.6 GB819details
RTX 508016 GBFitsF161.6 GB1170details
2x RTX 30902×24 GBFitsF161.6 GB1141details
2x RTX 40902×24 GBFitsF161.6 GB1229details
2x RTX 50902×32 GBFitsF161.6 GB2184details
RTX A600048 GBFitsF161.6 GB936details
RTX 6000 Ada48 GBFitsF161.6 GB1170details
RTX PRO 6000 Blackwell96 GBFitsF161.6 GB2184details
L40S48 GBFitsF161.6 GB1053details
A100 40GB40 GBFitsF161.6 GB1895details
A100 80GB80 GBFitsF161.6 GB2485details
H100 PCIe80 GBFitsF161.6 GB2438details
H100 SXM80 GBFitsF161.6 GB4083details
H200141 GBFitsF161.6 GB5851details
RX 76008 GBFitsF161.6 GB351details
RX 7700 XT12 GBFitsF161.6 GB527details
RX 7800 XT16 GBFitsF161.6 GB761details
RX 7900 GRE16 GBFitsF161.6 GB702details
RX 7900 XT20 GBFitsF161.6 GB975details
RX 907016 GBFitsF161.6 GB780details
RX 9070 XT16 GBFitsF161.6 GB780details
Ryzen AI Max+ 395 128GB128 GBFitsF161.6 GB312details
Arc A770 16GB16 GBFitsF161.6 GB683details
Mac M1 16GB16 GBFitsF161.6 GB83details
Mac M2 24GB24 GBFitsF161.6 GB122details
Mac M3 24GB24 GBFitsF161.6 GB122details
Mac M4 16GB16 GBFitsF161.6 GB146details
Mac M4 32GB32 GBFitsF161.6 GB146details
Mac M1 Pro 32GB32 GBFitsF161.6 GB244details
Mac M3 Pro 36GB36 GBFitsF161.6 GB183details
Mac M4 Pro 24GB24 GBFitsF161.6 GB333details
Mac M4 Pro 64GB64 GBFitsF161.6 GB333details
Mac M1 Max 64GB64 GBFitsF161.6 GB488details
Mac M2 Max 96GB96 GBFitsF161.6 GB488details
Mac M3 Max 128GB128 GBFitsF161.6 GB488details
Mac M4 Max 64GB64 GBFitsF161.6 GB666details
Mac M2 Ultra 192GB192 GBFitsF161.6 GB975details
Mac M3 Ultra 256GB256 GBFitsF161.6 GB998details

Questions

How much VRAM does SmolVLM2 500M Video Instruct need?

About 1.1 GB at 4-bit (Q4_K_M) with an 8K context: 0.3 GB of weights (estimated), 0.3 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 1.2 GB.

What is the smallest GPU that runs SmolVLM2 500M Video Instruct?

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 SmolVLM2 500M Video Instruct 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.

SmolVLM 500M Instruct

507M · 1.1 GB at 4-bit

SmolLM2 360M Instruct

362M · 1.1 GB at 4-bit

SmolLM2 360M

362M · 1.1 GB at 4-bit

Qwen2.5 0.5B Instruct

494M · 1.2 GB at 4-bit

GOT OCR 2.0 hf

561M · 1.8 GB at 4-bit

LFM2.5 VL 450M

449M · 0.8 GB at 4-bit

Model facts

Hugging Face repoHuggingFaceTB/SmolVLM2-500M-Video-Instruct
Parameters507,482,304 (507M)
Layers / hidden size32 / 960
AttentionGQA; 15 heads, 5 KV heads, head dim 64
KV cache per token (FP16)40.0 KiB
Max context8,192 tokens
Licenseapache-2.0
Released on HFFeb 11, 2025
Downloads (30d) / likes1,707,180 / 206

Sources: Hugging Face model API · config.json (HuggingFaceTB/SmolVLM2-500M-Video-Instruct) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.