SmolVLM2 2.2B Instruct
By HuggingFace · France
Updated 2026-07-13
Overview
HuggingFace's 2.2B vision-language model built on SmolLM2-1.7B, handling image, video, and text in roughly 5.2GB of VRAM. The smallest serious VLM with video understanding.
When to pick this model
- Adding vision to mobile or embedded apps
- Video frame analysis on a single consumer GPU
- Document and screenshot understanding at the edge
- Permissively licensed multimodal prototypes
- Bandwidth-constrained deployments needing local VLM
VRAM requirements by quantization
| Quantization | VRAM required |
|---|---|
| Q4_K_M (recommended) | 1.6 GB |
| Q5_K_M | 2 GB |
| Q8_0 | 3 GB |
| FP16 (no quantization) | 4.5 GB |
VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.
In practice, SmolVLM2 2.2B Instruct fits an 8 GB consumer card at Q4_K_M (1.6 GB). Stepping up to Q8_0 nearly doubles the footprint to 3 GB, and unquantized FP16 weights take 4.5 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.
Without a GPU, SmolVLM2 2.2B Instruct needs roughly 5 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 30 tokens/sec on entry-level GPUs, on the order of 90 tokens/sec on a mid-range card, and up to 180 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.
What hardware do you need
The table below matches SmolVLM2 2.2B Instruct to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.
| GPU memory | Example cards | Best fit for SmolVLM2 2.2B Instruct |
|---|---|---|
| 8 GB | RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB | FP16 (4.5 GB used) |
| 12 GB | RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop | FP16 (4.5 GB used) |
| 16 GB | RTX 5080, RTX 4080 Super, Radeon RX 9070 XT | FP16 (4.5 GB used) |
| 24 GB | RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop | FP16 (4.5 GB used) |
| 32 GB | RTX 5090 | FP16 (4.5 GB used) |
Which GPU should you buy to run SmolVLM2 2.2B Instruct?
To run SmolVLM2 2.2B Instruct locally at Q4, you need ~1.6 GB of VRAM. The best value for this is a RTX 5060 (8 GB VRAM).
As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.
Strengths
- Runs full video inference in ~5.2GB VRAM
- Apache 2.0 license suitable for commercial use
- Genuine image + video + text capability at 2.2B scale
- Inherits SmolLM2's tight text fundamentals
Limitations
- 8K context inherited from SmolLM2 limits long video
- No official Ollama distribution yet
- Video understanding is basic compared to frontier VLMs
Typical workloads
In our catalog grid, SmolVLM2 2.2B Instruct is filed under Ultra-Compact Vision, Basic Video — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: vision-language work — screenshots, charts, scanned documents.
Note the 8k-token context window — fine for short interactions, limiting for long documents or big retrieval contexts. The Apache 2.0 license is permissive, so shipping it inside a commercial product raises no special legal questions.
Architecture & training
Architecture: VLM image+video+text → text · SmolLM2-1.7B backbone
Training: ~5.2 GB VRAM for video inference.
The go-to small VLM when you need vision plus video in under 3B parameters and an Apache license.
Quick start
# HuggingFace : HuggingFaceTB/SmolVLM2-2.2B-InstructOr use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.
Similar models worth comparing
Frequently asked questions
How much VRAM does SmolVLM2 2.2B Instruct need?
At the recommended Q4_K_M quantization, SmolVLM2 2.2B Instruct needs about 1.6 GB of VRAM. Q8_0 takes 3 GB, and unquantized FP16 weights take 4.5 GB.
Can SmolVLM2 2.2B Instruct run without a GPU?
Yes — with roughly 5 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.
What context window does SmolVLM2 2.2B Instruct support?
SmolVLM2 2.2B Instruct supports a 8k-token context window (8,192 tokens).
Can I use SmolVLM2 2.2B Instruct commercially?
Yes. SmolVLM2 2.2B Instruct is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.
How fast is SmolVLM2 2.2B Instruct on consumer hardware?
Our compatibility engine estimates on the order of 90 tokens/sec on a mid-range GPU and up to 180 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.
Which quantization of SmolVLM2 2.2B Instruct should I download first?
Start with Q4_K_M (1.6 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It already fits an 8 GB card at FP16.