Chloe Model Specifications

2026-01-29 18:00 124 words 1 min read

no table of contents
Frontier AI. Own attention math. Own architecture. Coming soon.

Prior Art Declaration

Chloe model family specifications.


Model Sizes

FAMILY:
- 12GB
- 24GB
- 36GB
- 48GB
- 60GB (planned)
- 64GB (planned)

increment: 12GB
scaling: linear

Target Performance

PROJECTED:
- competitive with current frontier models
- reasoning, coding, instruction following

Resources

EFFICIENCY:
- designed for less compute
- designed for less memory
- designed for lower inference cost

more capability per resource uwu

Architecture

SPECIALIZED MoE:
- params: ~1 trillion
- experts: ~400
- architecture: proprietary MoE design
- routing: custom attention-based

Details in separate filings.


Record

Date: January 29, 2026

Author: rune.みんな

Status: Claimed


rune.みんな

© 2024 - 2026 rune.みんな
Powered by theme astro-koharu · Inspired by Shoka