YOU ARE VIEWING ONE ITEM FROM THE AICRIER FEED

FineFab releases 35B embedded engineering model micro-kiki-v3

AICrier tracks AI developer news across Product Hunt, GitHub, Hacker News, YouTube, X, arXiv, and more. This page keeps the article you opened front and center while giving you a path into the live feed.

// WHAT AICRIER DOES

7+

TRACKED FEEDS

24/7

SCRAPED FEED

Short summaries, external links, screenshots, relevance scoring, tags, and featured picks for AI builders.

FineFab releases 35B embedded engineering model micro-kiki-v3
OPEN LINK ↗
// 45d agoMODEL RELEASE

FineFab releases 35B embedded engineering model micro-kiki-v3

Built by L'Électron Rare for FineFab, micro-kiki-v3 is a locally runnable Qwen3.5-based model featuring a 35-domain LoRA router and cross-session memory. It is explicitly tuned on real-world hardware consulting data for embedded engineering workflows.

// ANALYSIS

Releasing a specialized MoE with 35 dynamically routed LoRAs trained on consumer hardware shows the power of hyper-niche local AI.

  • The "negotiator" architecture actively arbitrates conflicting outputs between active domain stacks
  • Its "Aeon memory" enables cross-session graph persistence, a critical feature for long-running hardware projects
  • The open-sourced dataset includes 50K real-world Claude CLI sessions from an embedded consulting firm
  • The entire MLX-based training pipeline is open-source, allowing reproduction on Mac Studio hardware
// TAGS
micro-kiki-v3llmfine-tuningopen-weightsagent

DISCOVERED

45d ago

2026-04-18

PUBLISHED

45d ago

2026-04-18

RELEVANCE

8/ 10

AUTHOR

Holiday_Poetry_5133