AnythingLLM

All-in-one Desktop AI assistant. Private RAG, any LLM, Agents, and more.

Open source:

[AnythingLLM @AnythingLLM]

AnythingLLM is now available as an installable desktop application for MacOS and Windows

https://t.co/pnMbczWEhx

[AnythingLLM @AnythingLLM]

As of Desktop v1.18.0 we now run models via llama.cpp instead of Ollama.

• Up to 2x faster inference, depending on your hardware
• A runtime install that’s ~13x smaller
• Day 0 support for new models

Blog: https://t.co/HAIVhgDgnS

[AnythingLLM @AnythingLLM]

AnythingLLM 1.17.0 is out. We overhauled the Meeting Assistant to run @NVIDIAAI Nemotron 3 Diarization + @moondreamai Parakeet Redux, on your device.

3x smaller download
70-200x real-time on your hardware (GPU or CPU)
Integrated into your knowledgebase.

https://t.co/YK7BU4g8Ts

[Tim Carambat @tcarambat]

The @AnythingLLM Android app went a year without updates. It was well due for a complete re-work. It's live now and it's amazing, token efficient, and practical.

It won't compete for a private jet reservation.
It won't get you banned from OpenTable.

It just does what you want, right on your phone, without the infrastructure of a small nation. AI for the everyday.

RAG, agentic search, reranking, doc gen and recurring tasks. Private, open-source, no vendor lock-in, use an SLM on-device or BYOK.

Next up: one connected local-first experience across desktop, mobile and beyond.

From desktop, to mobile, to more - all on your hardware and no vendor lock-in.

https://t.co/N7VEDIDuzj

[AnythingLLM @AnythingLLM]

Almost every prompt requires web search. We are happy to announce our partnership with @youdotcom for AnythingLLM.

Every user now will get accurate, fresh, and rich web-search with ZDR right out of the box. Server, Desktop, & Mobile.

Read more: https://t.co/0xe0JtVkH0

[AnythingLLM @AnythingLLM]

Tab leverages the memories and activity you already have built up in AnythingLLM - make it a better suggestion engine the more you use it and write.

Download today for Mac and Windows:
https://t.co/9jKidE6CA8

[AnythingLLM @AnythingLLM]

AnythingLLM now ships with an OS-wide writing assistant called Magic Tab. Simply clicking on an input can make a contextual suggestion based on what you are looking at as well as contextual typeahead suggestions.

[AnythingLLM @AnythingLLM]

Free to use, powered with on-device models and seamlessly works with our tools, agents, memories, and more for Mac and Windows devices.

This is a better way to dictate. Download AnythingLLM for desktop today at https://t.co/9jKidE6CA8

[AnythingLLM @AnythingLLM]

"Intelligent Dictation" is live in @AnythingLLM. Now your local models clean up your words and paste them at your mouse cursor, but also can even see your open windows and give you better outputs based on what you're doing and what you said.

[AnythingLLM @AnythingLLM]

AnythingLLM is an app layer for local models. We believe that a local model can feel like a cloud model with the right tools and guardrails.

We believe we can bring 80% of your daily inference tasks to the computer you are already on.

https://t.co/LffCBgBiJl

[AnythingLLM @AnythingLLM]

Today, we demoed @PrismML's Bonsai 8B model in @AnythingLLM desktop to see if the 1-bit mode arch is the revolution local AI needs.

Turns out, this is not an April fools joke. 1.1GB memory footprint for a whole 8B model with high accuracy

[AnythingLLM @AnythingLLM]

The Bonsai models scale from mobile to Desktop (1.7B, 4B, and 8B) so you can get dense intelligence on any hardware.

You can find the full collection (including MLX) here.
https://t.co/mPgpMEd7wS

[AnythingLLM @AnythingLLM]

Over the weeked we passed by 50,000 GitHub stars on AnythingLLM!

We are so lucky to have so many people inspired, using, forking, and contributing to our vision of an LLM agnostic tool suite you can run locally or for your organization.

⭐️ https://t.co/EAPunPQouf

[AnythingLLM @AnythingLLM]

If you have specific models, benchmarks, and other details you want us to try out, let me know, and we can publish those in a follow-up video.

GA for these devices is published to be on Wednesday, Oct 15th, so you can let us try out your builds prior to get ideas.

[AnythingLLM @AnythingLLM]

We are a lucky partner to bench the new @NVIDIA DGX Spark before it came out. We build AI tools literally every day, so we were eager to get our hands on this to kick the tires.

[AnythingLLM @AnythingLLM]

128GB unified memory, 20-core ARM CPU, 4.1TB of storage, GB10 Blackwell superchip, and more. If you are interested in local AI inference in a single self-contained compute device, our overview video is worth checking out @NVIDIA_AI_PC 👉 https://t.co/ugvr5yuThG

[NVIDIA RTX Spark @NVIDIARTXSpark]

Turn notes/textbooks into an AI tutor with @AnythingLLM on RTX AI PCs.

Flashcards, quizzes, solutions—fast, private, local. 👉 https://t.co/g94E4HuCrr

[AnythingLLM @AnythingLLM]

AnythingLLM + RTX = 🔥

Checkout the latest blog on NVIDIA NIM + AnythingLLM Desktop

[NVIDIA RTX Spark @NVIDIARTXSpark]

Accelerate local LLMs and advanced AI workflows in @AnythingLLM with NVIDIA RTX. 🖥️💬

⚡ Fast inference w/ NVIDIA NIM Microservices
📁 Local privacy
🔁 Easy to use UI

Learn more in the #RTXAIGarage blog 👉 https://t.co/BQSlAPP0av

[NVIDIA RTX Spark @NVIDIARTXSpark]

NVIDIA & Microsoft Advance Development on RTX AI PCs.🤝

TensorRT has been reimagined for RTX. Available via Microsoft’s new inference stack, Windows ML, they streamline development and unlock peak performance.

Explore #MSBuild & #COMPUTEX2025 news 👉 https://t.co/rdZ38RTlHX

[NVIDIA RTX Spark @NVIDIARTXSpark]

AI at your fingertips. ✨

NVIDIA NIM microservices arrive on RTX AI PCs & workstations making AI tool creation easier than ever.

Plus, Project G-Assist System Assistant expands PC AI abilities with a custom plugin builder.

#RTXAIGarage: https://t.co/tmcRK22UAx

[AnythingLLM @AnythingLLM]

We ❤️ LanceDB

[LanceDB @lancedb]

⚔️ 𝗠𝗲𝗲𝘁 𝘁𝗵𝗲 𝗞𝗻𝗶𝗴𝗵𝘁𝘀 𝗮𝘁 𝗟𝗮𝗻𝗰𝗲𝗹𝗼𝘁 𝗥𝗼𝘂𝗻𝗱 𝗧𝗮𝗯𝗹𝗲 - 𝗧𝗶𝗺 ⚔️

@tcarambat is a fellow @ycombinator alumni, and has never been shy about his enthusiasm for LanceDB. Being a loyal ambassador, his team ships @AnythingLLM with LanceDB as a default for a seamless end-user experience!