Pinned Tweet
And here it is! Everything I've been carefully crafting for months.
I'm incredibly proud of this app and all that it is capable of.
Please check it out and let me know what you think!
I'm working hard to make it the best native AI experience users can have on Windows, and I have lots more to come.
Osaurus 1.0 for Windows is here.
Run local models. Give agents tools. Turn repeat tasks into workflows.
A native WinUI app with projects, voice calls, and images made on your GPU.
Free in the Microsoft Store. Come build private AI on Windows with us.
osaurus.ai/windows
If AI can become a geopolitical weapon, why are we all renting our intelligence from a handful of companies?
You don't need a conspiracy theory to see the risk.
Your AI shouldn't be collateral damage in someone else's political fight.
Own your AI. Before someone else decides you can't.
Bombshell story. The spy narrative is collapsing already.
Is there clear & convincing evidence that Alice Huang was a Chinese spy?
- AG says no. (“no conclusions reached”).
- CIA says no. (“not yet convinced”).
- DNI says no (“not entirely accurate and over-torqued”).
- Ranking Democrat on House Intelligence (incoming Chairman) says no. (“today, at least, I don’t know”). This was immediately after he was briefed.
Who argues otherwise? Anonymous officials tell CNN “it strains credulity to believe that she was not acting as an agent of the Chinese government — simply by virtue of the fact that she is a successful Chinese businesswoman working with one of America’s most powerful companies.”
This is their proof after a year of investigating and surveillance? That she was a “successful Chinese businesswoman”?
At this point, the only crime that we know for sure happened is the leaking of the classified memo. Who did that and why?
An investigation should be opened.
Joshua the son of Nun retweeted
After today's release of Osaurus for Windows, just wanted to say...
nitter.cf/windowsdev/status/2121…
Explore on-device AI and agents on Windows at #MSIgnite, from on-device app experiences to trusted AI agents. 🪟🤖
🚀 THR888 | 4:05 PM PT
🚀 THR889 | 4:40 PM PT
📅 Nov. 17
Join us in San Francisco or online! Add both to your schedule: ignite.microsoft.com/en-US/s…
llama.cpp v1.0.0
github.com/ggml-org/llama.cp…
Joshua the son of Nun retweeted
Local AI Builders Friday # 4
The list is now 122! A lot of you nominated people under # 3, thank you ❤️
I added 11 more amazing builders this week:
1) @dhh is an OG! He is working to transform Omarchy into a Local AI distro. Starting with Omarchy 4.6 - it’ll try to select the optimal AI model for your hardware and lets you share AI models across your tailnet (ie use your DGX Spark to provide AI capabilities to your netbook)
2) @0xSero (@exolabs) got DeepSeek running entirely offline on his pack-of-portable-compute (that he literally carries around in a backpack!) 🤯 and is getting 200+ tok/s
3) @ItsmeAjayKV is doing his own Local AI R&D (kernels, quants, etc.) using just a 3090 and 64GB of RAM - and is hands-down one of the best open weight model news aggregators/accounts to follow
4) @christopherowen runs TP4 on 4 DGX Sparks cabled in a ring (no switch!) and implemented his own one-shot RoCE all-reduce algorithm getting 13-19 µs allreduce latencies (vs 75-86 µs for NCCL). Check out his repo with 279 experiments (he also logs his misses, with numbers!)
5) @tpae is building @OsaurusAI, free, open source (MIT) software for running local AI on your devices. Check out his Apple @ Work appearance this week (his pitch: “own your AI. At work too.”)
6) @dealignai (also working on Osaurus AI!) makes the case that since real tasks are input-heavy, using prompt processing on 4 DGX Sparks instead of a 512GB M5 Ultra makes sense (similar $)
7) @tayarndt and 8) @mikedoise are building Perspective Intelligence at Techopolis - an on-device AI chat assistant for Mac, iPad and iPhone. @tayarndt is also behind DictateMute - free, open source software that automatically mutes you on Discord when you’re dictating to Claude Code
9) @thelogicalgate uses his 128GB M5 Max MBP and 128GB RTX Spark to run local LLMs and is working on his own inference engine for RTX chips (to run on Windows on Arm)
10) @aijoey is running some of the frontier open models on just 2 DGX Sparks and documenting everything (configs, benchmarks, even building tools like Spark Doctor to help others get the most out of their Sparks)
11) @rickmanelius has 3 Macs + 2 DGX Sparks all networked together using Tailscale so he always has access to 3 or 4 local AI models (and can decide which one to use for different tasks)
Follow them and follow the full list here: nitter.cf/i/lists/21089799789857…
Who should be in # 5?
Local AI Builders Friday # 3
The list is now 111. I learn more from this list than from the rest of my timeline these days!
I added 7 more amazing builders to follow this week:
1) @niklaslenz_ai runs local AI at home on 4 DGX Sparks and an RTX 5090 and ranks what runs best on the machine you own. His test of Idol turned one image into a 3D model in 14.5 s on a 5090 and 57.4 s on a 16GB 5060 Ti. Now he's building a strategy game where local models fight each other on any hardware.
2) @plotarmordev is setting up a Mac and DGX Spark cluster at home and sharing what he learns. On one Spark, TensorFold took Qwen3.8 Flash Next from 75 to 95 tok/s vs vLLM at 2 users, with 34% faster prefill.
3) @brexton co-heads AI and compute infra at Cantor. In his view every founder he talked to in SF is working on owning their own hardware.
4) @javierbits builds local AI and security tools in public. He's testing LFM3-1.7B, and this morning he hosted an X Space teaching how to get started with local AI on whatever hardware you have. I caught part of it and it was great.
5) @summerAI_O is an 18 year old student (happy birthday!) in Japan training models on local + cloud. He posts daily breakdowns of new AI papers with the numbers and the caveats and yesterday he shipped Diffusion-Qwen3.8-27B, which fills in many tokens per pass instead of writing one at a time.
6) @NotASithLord is CTO of Hydra Host, which runs GPU clusters for people who want to own their own. His take is that context is the most valuable thing an agent has and memory is the bottleneck.
7) @Prajhan_026 trained HobbyLM-1B from scratch, a sparse MoE with ~305M active params, in 76 hours on 4 H200s. Now it does computer use fully offline at 0.26 s per decision, 13x faster than before. Deep into research.
Follow them and follow the full list here: nitter.cf/i/lists/21089799789857…
Who should be in #4?
It was pure, happy coincidence that @OsaurusAI shipped 1.0 the same day as llama.cpp shipped 1.0, but not a coincidence that it ships with day 0 support for it.
nitter.cf/OsaurusAI/status/21212…
nitter.cf/ggerganov/status/21212…
llama.cpp v1.0.0 is now available
huggingface.co/blog/ggml-org…
Three weeks from public beta to 1.0.
Native C#. No Electron.
@SonofNun built the Windows app Osaurus deserved.
Osaurus 1.0 for Windows is here.
Run local models. Give agents tools. Turn repeat tasks into workflows.
A native WinUI app with projects, voice calls, and images made on your GPU.
Free in the Microsoft Store. Come build private AI on Windows with us.
osaurus.ai/windows
Joshua the son of Nun retweeted
A journalist's sources should never have to touch someone else's server.
With Mycroft 2.0, they don't.
Open source, running on the investigator's own machine. Grateful to be the runtime underneath.
Introducing Mycroft 2.0, an investigator harness built on @OsaurusAI. It runs investigations with models on your computer and connects them to all of my tools.
Ask it a question and it can:
- search a wiki of your notes and sources, kept in OpenKnowledge
- pull the figures out of a document or an image with an extraction model
- use a catalogue of research and verification skills from collaborators
- tell you what your Scoutpost monitors found on the pages and profiles you track
- plan an approach with OSINT Navigator, like how to trace the directors behind a shell company
- run a full Spotlight investigation from a lead, with a swarm of research agents and a separate fact-checker or analyst
- file what it finds into your wiki securely once you confirm
The Osaurus team put in a lot of hard work to get this running and I'm really grateful to them. Your sensitive cases and material can now stay entirely on your computer, where small models do the work with you.
Mycroft is open-source, and Indicator Lab members can install the whole setup with our desktop installer.
Went on Apple @ Work to talk about why we're building Osaurus.
Short version: own your AI. At work too.
Apple @ Work Podcast: Osaurus wants your company’s AI to run on the Mac, not in the cloud 9to5mac.com/2026/11/10/apple…
Apple @ Work Podcast: Osaurus wants your company’s AI to run on the Mac, not in the cloud 9to5mac.com/2026/11/10/apple…
This is a great ad.
AI is becoming the most personal software we'll ever use.
The more it knows about us, the more important it becomes to trust who controls it.
That's a big part of why we're building Osaurus.
Your thoughts are nobody's business.
Joshua the son of Nun retweeted
Replying to @Alibaba_Qwen
I’m ready. My dinos in @OsaurusAI are awaiting their brain upgrade.
Building agents for 4B local models taught me something surprising.
The smaller the model, the better the harness has to be.
Less context. Fewer tools. Clearer instructions. One step at a time.
Turns out, designing for small models is just designing better agents.
Joshua the son of Nun retweeted
Hadamard trellis combo to make it run on 48gb macs without hitting that wired ram limit, still holds enough quality to be decent at most jobs. Would have done SFT/GPRO but seeing how qwen 4 arch is around the corner I didn’t wanna waste time with it
Joshua the son of Nun retweeted
Allosaurus is all you need.
I’ve been using it extensively for some of my agentic workflows and I’d certainly say that it’s a solid substitute for deepseek on Macs that can’t run it
qwen3.8-flash-next running on 48GB Macs. Seriously impressive work by @dealignai!
My goal with @OsaurusAI is to make it stupid-simple to run local models. I’m not fully there yet, but I’m working on it.
@windows needs an app that makes this easy. Right now it is just too complicated even for technical people.
Watched the latest @LinusTech video and they are running QWEN 3.8 27B with pipeline parallelism GGUF on LM studio Q4KM and complaining it isn’t fast.
Oh my eyes.
Please PLEASE someone who’s close to
him educate him on VLLM and Strata.
He could literally be getting 5x better speeds and better intelligence.
Sure the setup is complex but it’s a one time setup with a cloud agent.
Own your intelligence.
Buy that RAM.
Do something cool that would have been impossible for you just two years ago.
Own your intelligence.
Anyone that opposes this has something to sell you (perhaps your own intelligence back to you.)
Local models need to continue proliferating. The “absolute unit” massive models may be out of reach for now locally, so your best option is to, indeed, keep your options open. Own your harness, data, connections, runtime: your agents. Embrace open source.
3
136






















