inclusive10 min read
How we fit a 27B model on a 12 GB laptop GPU
In short Qwen3.8-27B goes from 53.8 GB of 16-bit weights to an 8.31 GB Pingala-8 file, 84.5% smaller (6.5x). It runs with every layer on one 12 GB RTX 5070 Ti Laptop GPU at 37.7 tokens/s (llama-bench, batch 1).
By Genrius AIDev.to AI

Read the full story
Originally published by Dev.to AI on Nov 1, 2026.
Comments
Sign in to join the discussion.