Can I run SmolVLM Instruct locally?
SmolVLM Instruct by HuggingFaceTB has 2.2B parameters. At 4-bit with an 8K context it needs about 3.0 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.
Image input needs the vision projector file too: about 0.6 GB more (ggml-org/SmolVLM-Instruct-GGUF), not included in the figures below, which are for text-only use.
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/SmolVLM-Instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx |
|---|---|---|---|
| 2-bit (Q2_K) | rough | 2.0 GB | 2.8 GB |
| 3-bit (Q3_K_M) | usable | 2.3 GB | 3.0 GB |
| 4-bit (Q4_K_M) • | good, most popular | 2.3 GB | 3.0 GB |
| 5-bit (Q5_K_M) | very good | 2.7 GB | 3.5 GB |
| 6-bit (Q6_K) | near lossless | 3.0 GB | 3.7 GB |
| 8-bit (Q8_0) • | effectively lossless | 3.0 GB | 3.8 GB |
| 16-bit (F16/BF16) • | original weights | 4.6 GB | 5.4 GB |
Which GPUs and Macs run SmolVLM 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).
| Hardware | Memory | Fits? | Best quant | Needs | Max tok/s (est.) | Details |
|---|---|---|---|---|---|---|
| RTX 3060 12GB | 12 GB | Fits | F16 | 5.4 GB | 99 | details |
| RTX 4060 Ti 16GB | 16 GB | Fits | F16 | 5.4 GB | 79 | details |
| RTX 3090 | 24 GB | Fits | F16 | 5.4 GB | 258 | details |
| RTX 4090 | 24 GB | Fits | F16 | 5.4 GB | 278 | details |
| RTX 5090 | 32 GB | Fits | F16 | 5.4 GB | 494 | details |
| RTX 5070 Ti | 16 GB | Fits | F16 | 5.4 GB | 247 | details |
| RX 7900 XTX | 24 GB | Fits | F16 | 5.4 GB | 265 | details |
| Mac M4 Pro 48GB | 48 GB | Fits | F16 | 5.4 GB | 75 | details |
| Mac M4 Max 128GB | 128 GB | Fits | F16 | 5.4 GB | 151 | details |
| Mac M3 Ultra 512GB | 512 GB | Fits | F16 | 5.4 GB | 226 | details |
| DGX Spark | 128 GB | Fits | F16 | 5.4 GB | 75 | details |
| RTX 4070 | 12 GB | Fits | F16 | 5.4 GB | 139 | details |
| RTX 5060 Ti 16GB | 16 GB | Fits | F16 | 5.4 GB | 124 | details |
| RTX 4060 | 8 GB | Fits | F16 | 5.4 GB | 75 | details |
| Arc B580 | 12 GB | Fits | F16 | 5.4 GB | 126 | details |
| RTX 3060 Ti | 8 GB | Fits | F16 | 5.4 GB | 124 | details |
| RTX 3070 | 8 GB | Fits | F16 | 5.4 GB | 124 | details |
| RTX 3080 10GB | 10 GB | Fits | F16 | 5.4 GB | 210 | details |
| RTX 3080 12GB | 12 GB | Fits | F16 | 5.4 GB | 252 | details |
| RTX 3090 Ti | 24 GB | Fits | F16 | 5.4 GB | 278 | details |
| RTX 4060 Ti 8GB | 8 GB | Fits | F16 | 5.4 GB | 79 | details |
| RTX 4070 Super | 12 GB | Fits | F16 | 5.4 GB | 139 | details |
| RTX 4070 Ti | 12 GB | Fits | F16 | 5.4 GB | 139 | details |
| RTX 4070 Ti Super | 16 GB | Fits | F16 | 5.4 GB | 185 | details |
| RTX 4080 | 16 GB | Fits | F16 | 5.4 GB | 198 | details |
| RTX 4080 Super | 16 GB | Fits | F16 | 5.4 GB | 203 | details |
| RTX 5060 | 8 GB | Fits | F16 | 5.4 GB | 124 | details |
| RTX 5070 | 12 GB | Fits | F16 | 5.4 GB | 185 | details |
| RTX 5080 | 16 GB | Fits | F16 | 5.4 GB | 265 | details |
| 2x RTX 3090 | 2×24 GB | Fits | F16 | 5.4 GB | 258 | details |
| 2x RTX 4090 | 2×24 GB | Fits | F16 | 5.4 GB | 278 | details |
| 2x RTX 5090 | 2×32 GB | Fits | F16 | 5.4 GB | 494 | details |
| RTX A6000 | 48 GB | Fits | F16 | 5.4 GB | 212 | details |
| RTX 6000 Ada | 48 GB | Fits | F16 | 5.4 GB | 265 | details |
| RTX PRO 6000 Blackwell | 96 GB | Fits | F16 | 5.4 GB | 494 | details |
| L40S | 48 GB | Fits | F16 | 5.4 GB | 238 | details |
| A100 40GB | 40 GB | Fits | F16 | 5.4 GB | 429 | details |
| A100 80GB | 80 GB | Fits | F16 | 5.4 GB | 562 | details |
| H100 PCIe | 80 GB | Fits | F16 | 5.4 GB | 552 | details |
| H100 SXM | 80 GB | Fits | F16 | 5.4 GB | 924 | details |
| H200 | 141 GB | Fits | F16 | 5.4 GB | 1324 | details |
| RX 7600 | 8 GB | Fits | F16 | 5.4 GB | 79 | details |
| RX 7700 XT | 12 GB | Fits | F16 | 5.4 GB | 119 | details |
| RX 7800 XT | 16 GB | Fits | F16 | 5.4 GB | 172 | details |
| RX 7900 GRE | 16 GB | Fits | F16 | 5.4 GB | 159 | details |
| RX 7900 XT | 20 GB | Fits | F16 | 5.4 GB | 221 | details |
| RX 9070 | 16 GB | Fits | F16 | 5.4 GB | 176 | details |
| RX 9070 XT | 16 GB | Fits | F16 | 5.4 GB | 176 | details |
| Ryzen AI Max+ 395 128GB | 128 GB | Fits | F16 | 5.4 GB | 71 | details |
| Arc A770 16GB | 16 GB | Fits | F16 | 5.4 GB | 154 | details |
| Mac M1 16GB | 16 GB | Fits | F16 | 5.4 GB | 19 | details |
| Mac M2 24GB | 24 GB | Fits | F16 | 5.4 GB | 28 | details |
| Mac M3 24GB | 24 GB | Fits | F16 | 5.4 GB | 28 | details |
| Mac M4 16GB | 16 GB | Fits | F16 | 5.4 GB | 33 | details |
| Mac M4 32GB | 32 GB | Fits | F16 | 5.4 GB | 33 | details |
| Mac M1 Pro 32GB | 32 GB | Fits | F16 | 5.4 GB | 55 | details |
| Mac M3 Pro 36GB | 36 GB | Fits | F16 | 5.4 GB | 41 | details |
| Mac M4 Pro 24GB | 24 GB | Fits | F16 | 5.4 GB | 75 | details |
| Mac M4 Pro 64GB | 64 GB | Fits | F16 | 5.4 GB | 75 | details |
| Mac M1 Max 64GB | 64 GB | Fits | F16 | 5.4 GB | 110 | details |
| Mac M2 Max 96GB | 96 GB | Fits | F16 | 5.4 GB | 110 | details |
| Mac M3 Max 128GB | 128 GB | Fits | F16 | 5.4 GB | 110 | details |
| Mac M4 Max 64GB | 64 GB | Fits | F16 | 5.4 GB | 151 | details |
| Mac M2 Ultra 192GB | 192 GB | Fits | F16 | 5.4 GB | 221 | details |
| Mac M3 Ultra 256GB | 256 GB | Fits | F16 | 5.4 GB | 226 | details |
Questions
How much VRAM does SmolVLM Instruct need?
About 3.0 GB at 4-bit (Q4_K_M) with an 8K context: 1.0 GB of weights (real GGUF file size), 1.5 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 3.8 GB.
What is the smallest GPU that runs SmolVLM 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 SmolVLM 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.
SmolLM2 1.7B Instruct
1.7B · 3.0 GB at 4-bitSmolLM2 1.7B
1.7B · 3.0 GB at 4-bitSmolLM 1.7B
1.7B · 3.0 GB at 4-bitQwen3.5 2B
2.3B · 2.3 GB at 4-bitCosmos Reason2 2B
2.4B · 2.8 GB at 4-bitWeVisOne VL 2B
2.4B · 2.8 GB at 4-bitEmbed this
Free badge and "Will it run on my GPU?" widget for model cards, READMEs and blogs. No cookies, no tracking. All options.
Copy a snippet
Markdown badge
[](https://runsonmy.com/model/smolvlm-instruct/) Widget (script)
<div class="runsonmy-widget" data-model="smolvlm-instruct"><a href="https://runsonmy.com/model/smolvlm-instruct/">Will SmolVLM Instruct run on my GPU? Check on RunsOnMy</a></div>
<script async src="https://runsonmy.com/embed.js"></script> Widget (iframe)
<iframe src="https://runsonmy.com/embed/m/smolvlm-instruct/" width="100%" height="220" style="max-width:440px;border:0" loading="lazy" title="Will SmolVLM Instruct run on my GPU? (RunsOnMy)"></iframe> Model facts
| Hugging Face repo | HuggingFaceTB/SmolVLM-Instruct |
| Parameters | 2,246,272,880 (2.2B) |
| Layers / hidden size | 24 / 2048 |
| Attention | MHA; 32 heads, 32 KV heads, head dim 64 |
| KV cache per token (FP16) | 192.0 KiB |
| Max context | 16,384 tokens |
| License | apache-2.0 |
| Released on HF | Nov 18, 2024 |
| Downloads (30d) / likes | 17,193 / 602 |
Sources: Hugging Face model API · config.json (HuggingFaceTB/SmolVLM-Instruct) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.