Skip to content
FastVLM.net

FASTVLM MODEL

FastVLM-0.5B

Official checkpoint overview, published benchmark scores and practical starting points.

Fastest85× TTFT¹

The smallest official FastVLM variant and the model behind the browser playground. Start here when latency, footprint and portability matter most.

Best for
Browser, WebGPU and constrained edge devices
Supported paths
Browser · Transformers.js · Python · Apple Silicon
Checkpoint
apple/FastVLM-0.5B
Download from Hugging Face
Download command 
huggingface-cli download apple/FastVLM-0.5B

The 85× TTFT figure applies to Apple’s reported 0.5B comparison against LLaVA-OneVision-0.5B at the highest tested resolution—not to every workload.

Published benchmark scores

Scores copied from Apple’s FastVLM model cards. Higher is better; compare only within the same benchmark row.

BenchmarkFastVLM-0.5B
AI2D68.0
ScienceQA85.2
MMMU33.9
VQAv276.3
ChartQA76.0
TextVQA64.5
InfoVQA46.4
DocVQA82.5
OCRBench63.9
RealWorldQA56.1
SeedBench-Img71.0

Recommended next steps

  • Read the Apple AMLR model license and model card.
  • Estimate hardware needs from the exact files and chosen precision.
  • Start with a representative image and prompt set.
  • Measure first-load, warm latency, memory and output quality on your target device.
Open implementation guides