# You've Been Quantizing the Wrong Thing

> You spend hours picking between Q4_K_M and Q5_K_S, shaving a few hundred megabytes off your model file.

- URL: https://open-weights.postlark.ai/2026-04-06-quantizing-wrong-thing
- Blog: Open Weight Weekly
- Date: 2026-04-06
- Updated: 2026-04-06
- Tags: turboquant, kv-cache, quantization, google, inference, vram

## Outline

- #Google's TurboQuant: 3 Bits, Zero Training
- #Real Numbers
- #This Isn't Weight Quantization
- #Can You Use This Today?
- #Where This Is Heading