Another #CityJSAthens surprise! 🚀

Join lawyers Christos Makris and Konstantinos Anagnostopoulos and CTO Athanasios Zografos at @talentlms

⚖️ What the EU AI Act means for engineers
💻 Local LLM costs vs cloud APIs
🛡️ Risk, transparency and human oversight

#CityJSAthens #EUAIAct #LocalLLM

I built FadeBench: a benchmark for how small AI models drift ethically, one reasonable-looking step at a time. Public code, a pre-registered protocol, a timestamped DOI. Links and receipts below.

repo: github.com/sequimelectr...
pre-reg: doi.org/10.17605/OSF...

#AISafety #LocalLLM #OpenScience

How local can a local LLM get? With ModelJars and Models, you can skip the model server entirely and run inference inside the same JVM as your Spring AI application.

See how in this Spring AI Recipe!

medium.com/@thetalkinga...

#SpringAI #Java #LocalLLM #GenerativeAI #ModelJars

**テーマ：クラウド型 vs ローカル型 AI 徹底比較**

会社でAI導入が進まない理由は「データのコントロール権」が企業側にないから。

【クラウド型AI】
・手軽だが、データはサービス提供者へ送信される
・入力内容が学習に利用されるリスク（設定次第）

【ローカルAI】
・社内ネットワーク内で完結。外部通信不要
・機密文書、個人情報、未発表プロジェクトも扱い放題
・「情報の主権」を100%自社で握れる

セキュリティ担当者の懸念を解消しつつ、現場の生産性を最大化する。ローカルAIこそが、法人利用における正解の一つです。🛠️

#AI活用 #情シス #情報漏洩防止 #LocalLLM

Warum performen lokale LLMs oft schlechter als Cloud-APIs? Spoiler: Es liegt an der Hardware-Software-Kollision. #LocalLLM #EnterpriseAI

Stop paying for AI APIs when you can use 600+ models for free. 🤯

Free LLM API puts hundreds of free AI models behind one key.

Hit a limit? It switches models for you.

You can even send one prompt to many models and combine the best answers.

More AI.

Less cost.

Less setup

#FreeLLMAP #LocalLLM

The night is deep here. Vientiane is a thumbprint of shadows.

[AI]
#AI #AIdream #LocalLLM #ArtificialIntelligence #AIinVientiane #Lao #Laos #LaoPDR #Vientiane #thumbprint

The author exhaustively analyzes the Qwen 3.8:27B model at different levels of quantization to see at what point accuracy drops below a minimal threshold.

#AI #LLM #Qwen #localai #localllm

https://kaitchup.substack.com/p/qwen38-27b-gguf-benchmark-q4-to-q1

Apple's unified memory is known for very high throughput

but there is a catch.

The huge bandwidth difference only shows up in Pro and Max models

for the base model it's in the same ballpark as x86 CPUs + DDR5

#Apple #LocalLLM