pmetal search

Search HuggingFace Hub with memory fit estimation for your hardware.

Search the HuggingFace Hub for models. Shows memory fit estimation based on your Apple Silicon hardware.

Usage

pmetal search <QUERY> [OPTIONS]

Examples

# Basic search
pmetal search "qwen 0.6b"

# Detailed view with memory estimates
pmetal search "qwen 0.6b" --detailed

# Download the first result that fits
pmetal search "qwen 0.6b" --detailed --download

# JSON output
pmetal search "llama 3" --limit 10 --json

Parameters

Parameter	Default	Description
query	required	Search query
--limit	15	Maximum number of results
--download	false	Download the first result that fits
--detailed	false	Show memory breakdown and training estimates
--json	false	Emit JSON

Output

The --detailed flag shows memory requirements and whether the model fits in your device’s unified memory, accounting for GPU/ANE allocation and training overhead.

See Also

• pmetal download — Download a model
• Supported Models — Compatible architectures