Opinionated hardware classification library for Python.

caste is a deliberately simple library that inspects a user’s machine and assigns it to a small, practical "hardware caste" (like Mini, User, or Developer). This helps you set sensible defaults for your application based on the user's hardware without manually parsing RAM, CPU, and GPU stats.

pip install caste

import caste

# Get a simple one-word label for the machine
print(caste.detect_caste_word())  # e.g., "User"

# Get the label and the reason for the classification
label, reason = caste.detect_caste()
print(f"Caste: {label} (Reason: {reason})")

# Get raw hardware facts if you want to perform your own logic
facts = caste.detect_hw_facts()
print(f"RAM: {facts['ram_bytes'] / 1024**3:.1f} GB")
print(f"Discrete GPU: {facts['has_discrete_gpu']}")

• Mini — Microcomputers, embedded systems, classic/legacy PCs.
• User — Standard consumer-level machines.
• Developer — High-end personal computers.
• Workstation — Professional content-creation or high-end gaming machines.
• Rig — Dedicated servers or specialised compute systems.

detect_hw_facts() returns a dictionary with:

• ram_bytes (int)
• physical_cores (int)
• logical_threads (int)
• gpu_kind (0=None, 1=Integrated, 2=Unified, 3=Discrete)
• vram_bytes (int)
• has_discrete_gpu (bool)
• is_apple_silicon (bool)
• is_intel_arc (bool)

Modern applications often launch with no idea of the underlying hardware capability. If defaults are tuned only for a developer’s machine, it can lead to sluggish experiences on low-end hardware or not showing off your apps's best capabilities on high-end systems. caste provides a standardized way to bridge this gap.

There is also a C++ library and an executable for shell scripts and build systems. See the Github repo for details.

MIT