SmolVLM2-256M-Video-Instruct

Try Try

SmolVLM2-256M-Video-Instruct has 256M parameters (142M active per token) and takes 1.03 GB on disk. See where its parameters live, layer by layer, and what hardware it needs.

At a glance

HuggingFaceTB/SmolVLM2-256M-Video-Instruct
Repo createdFeb 11, 2025updated Nov 13, 2026
Model typeDenseFull attention (GQA)
InputsText + images86.4M vision encoder
Total parameters256M
Active per token142M55% of the model
ExpertsNone (dense)
Max context (from config)8K tokens
Layers30
On disk1.03 GB1 file
PrecisionFP32 (100%)
QuantizationNoneoriginal precision (FP32)
Fewest GPUs1× RTX 4060a single 8 GB card · weights only
FamilyQuantization of HuggingFaceTB/SmolVLM-256M-Instruct
Licenseapache-2.0
GitHubNot linked