Skip to content
Bitpute
ToolsModelsGPUsCloudLearn
Learn · Concepts

FP16 vs Q8 vs Q6 vs Q4: which quantization should you use?

Learn

By Bitpute · Published 12 July 2026 · Updated 24 July 2026 · How we estimate · Sources · Editorial policy · Version history · Report an error

On this page

You've picked a model. Now which build do you download? This is the practical decision — and for most people the answer is one line, with a few exceptions worth knowing.

Concept guideLast reviewed: July 2026

Every quantization is a point on one line: more bits = more quality and more VRAM; fewer bits = smaller and faster. New here? Start with what quantization is. Otherwise, here's how the common formats actually compare.

1The formats at a glance

FormatBits / weightQualityUse when
FP1616.0ReferenceYou have VRAM to burn or need exact reproducibility
Q8_08.5Near-losslessYou want maximum quality that still halves the size
Q6_K6.56ExcellentQuality-first with real VRAM savings
Q5_K_M5.69Very goodA safe margin above the sweet spot
Q4_K_M4.85The sweet spotBest model that fits — the default for most people
Q4_04.5GoodOlder/simpler; a touch more loss than Q4_K_M

2What that means in real VRAM

Same model, six builds. Here's Qwen2.5 14B (≈14.8B params), weights only — overhead and KV cache come on top (how that's calculated):

FormatWeightsComfortable card
FP16≈ 27.6 GiBA100 / dual 24 GB
Q8_0≈ 14.6 GiBRTX 4090 24 GB
Q6_K≈ 11.3 GiBRTX 4070 Ti SUPER 16 GB
Q5_K_M≈ 9.8 GiB16 GB card
Q4_K_M≈ 8.4 GiB12 GB card (RTX 3060/4070)
Q4_0≈ 7.8 GiB12 GB card

Cutting from FP16 to Q4_K_M drops the same model from ~27.6 GiB to ~8.4 GiB — a 14B model that needed a 24 GB card now fits on a 12 GB one.

3Which one — by what you're doing

Want the best model that fits → Q4_K_M

This is the right default. The VRAM you save lets you step up to a bigger, smarter model, and that almost always wins: a 14B at Q4_K_M beats a 7B at Q8 for the same memory. Optimize for model size first, precision second — down to Q4.

Want maximum quality and have the headroom → Q8_0

Effectively indistinguishable from FP16 at half the size. Reach for it when you're precision-sensitive (structured output, code that must compile, evals) and the model already fits comfortably.

Somewhere in between → Q5_K_M or Q6_K

A sensible margin when Q4 makes you nervous but Q8 won't fit. Little to fault here.

Tight on VRAM → stay at Q4_K_M, shrink the model

Don't drop below Q4 to squeeze a model in — the quality falls off a cliff there. Pick a smaller model at Q4_K_M instead. Use the model finder to see what fits your card.

4Bonus: lower quant is also faster

Decoding is memory-bandwidth bound — fewer bytes per weight means fewer bytes to read per token, so lower quants generate faster on the same card. The decode-speed ceiling is literally bandwidth ÷ weight-bytes, so Q4 has a higher ceiling than Q8 on identical hardware.

Size your model → What is quantization? Find a model that fits

Related: what VRAM is.

5Common questions

Should I use FP16 or Q4 for local inference?

Q4_K_M for almost all local single-user inference: it cuts weights by roughly 3.3× against FP16 for a small quality cost. FP16 is worth it when you have memory to spare and need maximum fidelity, or when a runtime requires it.

How much VRAM does quantization save?

Roughly in proportion to bits per weight. Moving from FP16 (16 bits) to Q4_K_M (about 4.85) cuts the weight term by about 3.3× — usually the single largest lever available.

Is a lower quantization also faster?

Yes, for a memory-bound workload. Decode speed is bounded by memory bandwidth divided by the model's size in memory, so a smaller model is read faster per token.