Is FlatQuant superseded?

FlatQuant (LLM quantization): superseded — cited as a baseline and beaten by newer methods. 2 paper(s) critique it, 4 beat it on benchmarks — #12 of 80 most-superseded. Sub-problem: cluster led by SmoothQuant. Newer alternatives in the same sub-problem include OffQ, FAIR-Calib, InfoQuant, ConQuR, Activation Residual Hessian Quantization (ARHQ).

LLM quantization

FlatQuant

FlatQuant: Flatness Matters for LLM Quantization

Superseded baseline#12 of 80 most-superseded · first seen Oct 12, 2024

Superseded — cited as a baseline and beaten by newer methods

2 papers critique it · 4 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites FlatQuant as a baseline.

FlatQuant flatquant demonstrates a loss of only 1.4 points

— OSC: Hardware Efficient W4A4 Quantization via Outlier Separation in Channel Dimension

“Even though these approaches successfully quantize LLMs to 4 bits with slight performance degradation, they apply the same type of transformation across all layers, ignoring the distribution characteristics of each layer within LLMs.”

— Adaptive Layer-Wise Transformations for Post-Training Quantization of Large Language Models

Beaten on benchmarks

Head-to-head results where a newer method reports beating FlatQuant. Values are copied from the source paper's tables — verify against the cited paper.

Reasoning-QAT beats FlatQuant
21.44 vs 17.34
Avg. · [Qwen3-0.6B W4A4KV4]
What Makes Low-Bit Quantization-Aware Training Work for Reasoning LLMs? A Systematic Study
FAIR-Calib beats FlatQuant
66.66 vs 63.98
Avg. · [Dream-Instruct]
FAIR-Calib: Frontier-Aware Instability-Reweighted Calibration for Post-Training Quantization of Diffusion Large Language Models
SliderQuant beats FlatQuant
5.71 vs 5.79
WikiText2 · [W4A4]
SliderQuant: Accurate Post-Training Quantization for LLMs

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.