Germain Masse

Debian Linux was officially founded by Ian Murdock on August 16, 1993. That's 32 years of stable, free, and opensource computing. From a single kernel, it has grown to parent some of the most popular OSes in the world. Happy 32nd birthday to the beloved mother of all distros! 🥰 🍰

The Big LLM Architecture Comparison From DeepSeek-V3 to Kimi K2: A Look At Modern LLM Architecture Design

The Big LLM Architecture Comparison, by the excellent @sebastianraschka.com

magazine.sebastianraschka.com/p/the-big-ll...

AI coding tools are shifting to a surprising place: the terminal | TechCrunch For years, code-editing tools like Cursor, Windsurf, and GitHub’s Copilot have been the standard for AI-powered software development. But as agentic AI grows more powerful and vibe-coding takes off, a subtle shift has changed how AI systems are interacting with software. Instead of working on code, they’re increasingly interacting directly with the shell of whatever system they’re installed in. It’s a significant change in how AI-powered software development happens – and despite the low profile, it could have significant implications for where the field goes from here.

AI coding tools are shifting to a surprising place: the terminal

Moonshot AI just released Kimi K2, an open-source 1T MoE LLM with 32B active params. SOTA on coding, reasoning, agents. API and weights are live.
moonshotai.github.io/Kimi-K2/

Let's Encrypt rolls out free security certs for IP addresses You probably don't need one, but it's nice to have the option Let's Encrypt, a certificate authority (CA) known for its free TLS/SSL certificates, has begun issuing digital certificates for IP addresses.…

Let's Encrypt rolls out free security certs for IP addresses

Using Function Calling with OVHcloud AI Endpoints - OVHcloud Blog If you want to have more information on AI Endpoints, please read the following blog post.You can, also, have a look at our previous blog posts on how use AI Endpoints. OVHcloud AI Endpoints allows de...

Je suis joie 🥳
AI Endpoints by OVHcloud est compatible avec le function calling 🤝.

Pour le voir en action avec @langchain4j cela se passe ici : blog.ovhcloud.com/using-functi...

Et fin de semaine un nouveau blog post pour le ... 🤫

Curation is the New Discovery API modesty for the agentic era.

For years, we've designed APIs for maximum discoverability. The more features a developer can find, the better.

In an agentic world, that entire philosophy is a liability. The best API for an agent is the one it can't see.

Even more on designing products for agents: www.jlowin.dev/blog/curatio...

Today's LLMs craft exploits from patches at lightning speed Erlang? Er, man, no problem. ChatGPT, Claude to go from flaw disclosure to actual attack code in hours The time from vulnerability disclosure to proof-of-concept (PoC) exploit code can now be as short as a few hours, thanks to generative AI models.…

Today's LLMs craft exploits from patches at lightning speed

Et vous, c'est quoi votre pire cauchemar ?

🚨 LLM Alert: Meet DeepDuck-70B, the first LLM trained to speak fluent duck. 🦆
Trained on 5T+ quacks, it’s redefining avian NLP.

Try it now on AI Endpoints: c1ffa52d-a184-4b7b-9824-7b8033d7bfbd.app.gra.ai.cloud.ovh.net

🔍 More models: endpoints.ai.cloud.ovh.net

#AI #LLM #QuackToTheFuture #OVHcloud

LLM providers on the cusp of an 'extinction' phase as capex realities bite Only the strong will survive, but analyst says cull will not be as rapid as during dotcom era Gartner says the market for large language model (LLM) providers is on the cusp of an extinction phase as it grapples with the capital-intensive costs of building products in a competitive market.…

LLM providers on the cusp of an 'extinction' phase as capex realities bite

« Intel’s former CEO Pat Gelsinger claims NVIDIA’s AI GPUs are 10,000 times more expensive than what is needed for AI inferencing »
https://wccftech.com/intel-former-ceo-claims-nvidia-ai-gpus-are-10000-times-more-expensive-than-what-is-needed-for-ai-inferencing/

Pruna AI open sources its AI model optimization framework Pruna AI, a European startup that has been working on compression algorithms for AI models, is making its optimization framework open source on Thursday. Pruna AI has been creating a framework that applies several efficiency methods, such as caching,…

Pruna AI open sources its AI model optimization framework

On my way to NVIDIA #GTC25!
If you’re there, let’s connect!

Large enterprises scramble after supply-chain attack spills their secrets tj-actions/changed-files corrupted to run credential-stealing memory scraper.

The latest open source supply-chain attack to roil the Internet targets software used by more than 23,000 organizations, some of them in large enterprises.

GitHub - camel-ai/camel: 🐫 CAMEL: Finding the Scaling Law of Agents. The first and the best multi-agent framework. https://www.camel-ai.org 🐫 CAMEL: Finding the Scaling Law of Agents. The first and the best multi-agent framework. https://www.camel-ai.org - camel-ai/camel

🚀 Skyrocketing！ 🚀 (200+ new stars)

📦 camel-ai / camel
⭐ 7,253 (+350)
🗒 Python

🐫 CAMEL: Finding the Scaling Law of Agents. The first and the best multi-agent framework. https://www.camel-ai.org

On Ukraine, Europe must step into the void | Virgin

Europe can’t rely on the strength of the transatlantic relationship to help contain Russia’s threat – it must lead from the front and be the true force of support that Ukraine needs. Not just for Ukraine’s sake, but for a free and sovereign Europe.

Nearly 10 years after Data and Goliath, Bruce Schneier says: Privacy’s still screwed 'In 50 years, I think we'll view these business practices like we view sweatshops today' Interview  It has been nearly a decade since famed cryptographer and privacy expert Bruce Schneier released the book Data and Goliath: The Hidden Battles to…

Nearly 10 years after Data and Goliath, Bruce Schneier says: Privacy’s still screwed

Bypass DeepSeek censorship by speaking in hex (substack.com)

Main Link | Discussion

Deepseek-R1 distilled into Llama-70B is now live on #OVHcloud AI Endpoints!
🚀 Try it for free during the beta!
https://endpoints.ai.cloud.ovh.net

The Illustrated DeepSeek-R1 (newsletter.languagemodels.co)

Main Link | Discussion

ARM-based Instances · Issue #691 · ovh/public-cloud-roadmap As an OVHcloud customer using compute instances, I want to have ARM-based instances to optimize Performance, Energy consumptions and potentially spendings.

My OVHcloud colleagues are currently evaluating a few options to offer #ARM instances in #ovhcloud 's public cloud compute catalog.
We are really interested in the ideal ARM flavors in terms of core/ vs ram vs disk vs bandwith. Please comment this issue : github.com/ovh/public-c... #iaas #cloud

GitHub - Dokploy/dokploy: Open Source Alternative to Vercel, Netlify and Heroku. Open Source Alternative to Vercel, Netlify and Heroku. - Dokploy/dokploy

📦 Dokploy / dokploy
⭐ 11,163 (+89)
🗒 TypeScript

Open Source Alternative to Vercel, Netlify and Heroku.

GitHub - serengil/deepface: A Lightweight Face Recognition and Facial Attribute Analysis (Age, Gender, Emotion and Race) Library for Python A Lightweight Face Recognition and Facial Attribute Analysis (Age, Gender, Emotion and Race) Library for Python - serengil/deepface

🚀 Skyrocketing！ 🚀 (200+ new stars)

📦 serengil / deepface
⭐ 16,146 (+405)
🗒 Python

A Lightweight Face Recognition and Facial Attribute Analysis (Age, Gender, Emotion and Race) Library for Python

GitHub - Zipstack/unstract: No-code LLM Platform to launch APIs and ETL Pipelines to structure unstructured documents No-code LLM Platform to launch APIs and ETL Pipelines to structure unstructured documents - Zipstack/unstract

📦 Zipstack / unstract
⭐ 2,624 (+25)
🗒 Python

No-code LLM Platform to launch APIs and ETL Pipelines to structure unstructured documents

GitHub - 1Panel-dev/1Panel: 🔥 Top-Rated Web-Based Linux Server Management Tool. 1Panel features an intuitive web interface that seamlessly integrates server management and monitoring, container management, database administration, website management, system backup and restoration, and more. 🔥 Top-Rated Web-Based Linux Server Management Tool. 1Panel features an intuitive web interface that seamlessly integrates server management and monitoring, container management, database administra...

📦 1Panel-dev / 1Panel
⭐ 23,678 (+67)
🗒 Go

🔥 Top-Rated Web-Based Linux Server Management Tool. 1Panel features an intuitive web interface that seamlessly integrates server management and monitoring, container management, database administration, website management, system backup and res...

New LLM optimization technique slashes memory costs (venturebeat.com)

Main Link | Discussion

Am I the only one struggling to find detailed GPU specs? That’s why I started this project. If you know of other data sources, feel free to share! https://github.com/gmasse/gpu-specs

GitHub - langflow-ai/langflow: Langflow is a low-code app builder for RAG and multi-agent AI applications. It’s Python-based and agnostic to any model, API, or database. Langflow is a low-code app builder for RAG and multi-agent AI applications. It’s Python-based and agnostic to any model, API, or database. - langflow-ai/langflow

🔥 Hot Repo！ 🔥 (100+ new stars)

📦 langflow-ai / langflow
⭐ 37,757 (+186)
🗒 Python

Langflow is a low-code app builder for RAG and multi-agent AI applications. It’s Python-based and agnostic to any model, API, or database.