FastVLM 0.5B

FastVLM 0.5B needs an estimated 1 GB of VRAM at Q4, 1.4 GB at Q8 and 2.2 GB at FP16. All figures are estimates from the 759M parameter count, covering weights plus about 15% and 0.5 GB for a short context.

Yes, it fits on all of them. At FP16 it needs an estimated 2.2 GB, so it fits at FP16 on a 24 GB, 16 GB, 12 GB or 8 GB GPU.

Yes. Q4 needs an estimated 1 GB of VRAM, and assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is an estimated 8 GB.

Yes, you can self-host it. FastVLM 0.5B has open weights, so you can run it yourself for full control of your data.

Run FastVLM 0.5B locally

It reads text and images. Audio, video and PDFs are not listed as input modalities.

Matched on model kind and release date, with no comparable price, the closest alternatives are Llama Prompt Guard 2 22M and Llama Prompt Guard 2 86M by Meta, and Qwen3 8B by Alibaba.

Llama Prompt Guard 2 22MFastVLM 0.5B vs Llama Prompt Guard 2 22MLlama Prompt Guard 2 86MFastVLM 0.5B vs Llama Prompt Guard 2 86MQwen3 8BFastVLM 0.5B vs Qwen3 8B

It has open weights, not fully open source. The license is apple-amlr.

You can download the weights from the Hugging Face repository apple/FastVLM-0.5B.

Run FastVLM 0.5B locally

FastVLM 0.5B was made by Apple and released on May 7, 2025.

More from Apple

FastVLM 0.5B has open weights. The listed facts do not state pricing, so whether it is free to use is not specified.