Specification

Release date: 11 Feb 2025
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Openness: open-weights
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
License: apache-2.0
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Architecture: SmolVLMForConditionalGeneration
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Parameters: 256.5M
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Languages: en
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Base model: HuggingFaceTB/SmolVLM-256M-Instruct
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Quantization format: onnx
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
File size: 1.0 GB
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Hugging Face repo: HuggingFaceTB/SmolVLM2-256M-Video-Instruct
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Pipeline tag: image-text-to-text
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Model card: huggingface.co/HuggingFaceTB/SmolVLM2-256M-Video-Instruct
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Downloads: 50,818
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Likes: 114
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Datasets: HuggingFaceM4/the_cauldron, HuggingFaceM4/Docmatix, lmms-lab/LLaVA-OneVision-Data, lmms-lab/M4-Instruct-Data, HuggingFaceFV/finevideo, MAmmoTH-VL/MAmmoTH-VL-Instruct-12M, lmms-lab/LLaVA-Video-178K, orrzohar/Video-STaR, Mutonix/Vript, TIGER-Lab/VISTA-400K, Enxin/MovieChat-1K_train, ShareGPT4Video/ShareGPT4Video
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Gated: No
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Is quantized: Yes
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Last modified: 2025-04-08T07:25:59+00:00
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Library name: transformers
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Downloads all time: 2,186,053
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Model type: smolvlm
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Tags: transformers, onnx, safetensors, smolvlm, image-text-to-text, en
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium
Weights dtype: F32
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium

Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →

Provenance

Attributed facts

23

Source tiers

T223

Freshest observation

43 min ago

Conflicts

None

Relations

Described by: arXiv:2504.05299
Developed by: Hugging Face
Quantized from: SmolVLM-256M-Instruct
Trained on: lmms-lab/M4-Instruct-Data, HuggingFaceFV/finevideo, MAmmoTH-VL/MAmmoTH-VL-Instruct-12M, lmms-lab/LLaVA-Video-178K, HuggingFaceM4/the_cauldron, Mutonix/Vript, TIGER-Lab/VISTA-400K, Enxin/MovieChat-1K_train +4(12 total)

Identity

hf_repo: HuggingFaceTB/SmolVLM2-256M-Video-Instruct

Also known as: HuggingFaceTB/SmolVLM2-256M-Video-Instruct, SmolVLM2-256M-Video-Instruct

slug smolvlm2-256m-video-instruct

ModelSmolVLM2-500M-Video-InstructHugging Face
ModelSmolVLM2-2.2B-InstructHugging Face
ModelSmolVLM-InstructHugging Face
ModelSmolVLM-500M-InstructHugging Face
ModelSmolVLM-256M-InstructHugging Face
ModelSmolLM-1.7B-InstructHugging Face
ModelSmolLM-360M-InstructHugging Face
Modelstack-edu-classifier-pythonHugging Face

Compare

Open compare →

Modalities

Modalities unavailable.

Capabilities

Tool calling
Unavailable
Structured output
Unavailable
Reasoning
Unavailable
Vision
Unavailable
Audio
Unavailable
Fine-tuning available
Unavailable

No capability flags have been observed from a source yet — we do not infer them.

Languages: en
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 43 min agomedium

No benchmark results recorded

Results appear when a tier 1–3 source publishes them; we never copy scores without a source.

Current prices

No current prices recorded

Prices appear when a provider publishes a public pricing page we crawl.

Price history

Estimated

Memory need = bytes per parameter (4-bit ≈ 0.5 × 1.15 overhead, 8-bit 1.0, fp16 2.0) + a KV-cache allowance. Not a measurement.

Estimated hardware fit
Hardware	Quantization	Memory	Est. need	Fits
Apple M3 Ultra	4bit	—	0.7 GB	Yes
AMD Instinct MI325X	4bit	256 GB	0.7 GB	Yes
AMD Instinct MI300X	4bit	192 GB	0.7 GB	Yes
Apple M2 Ultra	4bit	—	0.7 GB	Yes
NVIDIA B200	4bit	180 GB	0.7 GB	Yes
NVIDIA H200	4bit	141 GB	0.7 GB	Yes
Apple M1 Ultra	4bit	—	0.7 GB	Yes
Apple M3 Max	4bit	—	0.7 GB	Yes
Apple M4 Max	4bit	—	0.7 GB	Yes
NVIDIA DGX Spark	4bit	128 GB	0.7 GB	Yes
Apple M2 Max	4bit	—	0.7 GB	Yes
NVIDIA A100 80GB	4bit	80 GB	0.7 GB	Yes
NVIDIA H100 SXM	4bit	80 GB	0.7 GB	Yes
Apple M1 Max	4bit	—	0.7 GB	Yes
Apple M4 Pro	4bit	—	0.7 GB	Yes
Apple M3 Pro	4bit	—	0.7 GB	Yes
Apple M1 Pro	4bit	—	0.7 GB	Yes
Apple M2 Pro	4bit	—	0.7 GB	Yes
Apple M4	4bit	—	0.7 GB	Yes
NVIDIA GeForce RTX 5090	4bit	32 GB	0.7 GB	Yes
Apple M2	4bit	—	0.7 GB	Yes
Apple M3	4bit	—	0.7 GB	Yes
NVIDIA GeForce RTX 3090	4bit	24 GB	0.7 GB	Yes
NVIDIA GeForce RTX 4090	4bit	24 GB	0.7 GB	Yes
Apple M1	4bit	—	0.7 GB	Yes

Ancestors 4

This model

SmolVLM2-256M-Video-Instruct

256.5M params

Quantizations 0

None recorded.

Descendants 0

None recorded.

Papers 1

arXiv:2504.05299Active

Repositories 0

No repositories linked yet.

As of

Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.

Claim history · Datasets

1 claims · 1 propertiesShow all properties

Datasetsdatasets1

Claim history for Datasets
Value	Valid from → to	Status	Source	Confidence	Extractor
HuggingFaceM4/the_cauldron, HuggingFaceM4/Docmatix, lmms-lab/LLaVA-OneVision-Data, lmms-lab/M4-Instruct-Data, HuggingFaceFV/finevideo, MAmmoTH-VL/MAmmoTH-VL-Instruct-12M, lmms-lab/LLaVA-Video-178K, orrzohar/Video-STaR, Mutonix/Vript, TIGER-Lab/VISTA-400K, Enxin/MovieChat-1K_train, ShareGPT4Video/ShareGPT4Video	11 Sept 2026 21:35 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

11 Sept 2026
New modelModelSmolVLM2-256M-Video-InstructHugging Face
New model: SmolVLM2-256M-Video-Instruct (Hugging Face)
huggingface
huggingface.co

Full timeline →

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/HuggingFaceTB/SmolVLM2-256M-Video-Instruct	model_page	T2· Quality secondary	40 min ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/models?author=HuggingFaceTB&p=0&sort=downloads	listing	T2· Quality secondary	1 h ago	2

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.