Jaehoon Lee

Technical Content Manager

Insight

Ternary quantization results: model size drops from 60GB in BF16 to about 10GB at 1.58 bits (about 1/6), with performance at 89% of BF16
Tech
November 10, 2026
|
7

min read

Can a 30B Local LLM Run on a MacBook? 1.58-Bit Ternary Quantization

Nota AI Server runs the same LLM on 4 GPUs, saving 12 GPUs compared with an original server that needs 16
Product
October 21, 2026
|
16

min read

[Nota AI Server] Building an On-Premises LLM: What to Check Before You Buy GPUs

A banner stating a 7x speedup, with inference latency cut from 1.6 seconds to 230ms. The figure comes from running GR00T N1.7 3B on a Qualcomm IQ-9075.
Tech
September 18, 2026
|
14

min read

Running VLA Without a GPU: 1.6s to 230ms on a Qualcomm NPU

Graphic showing GPU reduction for Solar Open2 and Kimi K3. A stack of 8 GPUs shrinks to 2 for Solar Open2 and to 4 for Kimi K3.
Tech
August 26, 2026
|
7

min read

From Eight H100s to Two: The GPU Cost Savings of LLM Compression

Thumbnail showing the NetsPresso logo and a capture of the agent interface over a purple-to-blue gradient. The center shows a "Welcome to NetsPresso!" message inviting the user to describe the model to optimize and where it needs to run, followed by a high
Product
August 5, 2026
|
3

min read

From Semi-Automatic to Fully Automatic: How NetsPresso Optimizes Models with an Agent

Events
July 30, 2026
|
4

min read

Nota AI Joins AMD's Robotics Partner Network as the Only Korean Company

Culture

Team Nota AI
September 1, 2026
|
8

min read

What Six Months at Nota AI Looks Like

Stay Ahead with Expert Insights on Edge AI

Subscribe Now
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Site Map
Insights
Culture
Newsroom
Family Site
Nota AI®NetsPresso®Careers
Owned Media
LinkedinYoutubeFacebookX
© 2021-2026. Nota Inc. All rights reserved.
Privacy PolicyContact Us