@tpae
Account based inUnited States
About this account
- Account based in
- United States
- Connected via
- United States App Store
Account-level information from X, not a live location or the device used for a specific post.
building @OsaurusAI - prev @Tesla @Netflix
California, USA
Joined April 2009
- Tweets2.5K
- Following364
- Followers1.2K
- Likes7.8K
tpae retweeted
The main appeal of open-source agent harnesses isn't that they are free, but that we can inspect what they are doing on our computers.
tpae retweeted
Most local AI demos run on maxed-out Macs.
40% of our users have 16GB or less.
Raptor is for them. 4B. Wi-Fi off.
As much as I think Muse is great, people forget that Mark Zuckerberg, the ultimate extractor of people's privacy, now has even more access than ever. Don't be surprised when Muse starts suggesting products and services and paying for them on your behalf.
Privacy needs constant education and awareness to survive.
Don’t make the same mistake.
threads.com/share/BABlDQUr58…
The part that got me is it noticed the $2,800 ad spend was eating the whole margin. 🤯
tpae retweeted
Bonsai 2 dropped hours ago. It's already doing real work in Osaurus.
Read two CSVs. Wrote the Python. Drew the chart.
Then found the one channel losing money and told me why.
27B. Ternary. Nothing left the Mac.
Today, we’re announcing Ternary Bonsai 2 27B.
Based on Qwen3.8 27B, Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance.
Two months after the first Bonsai 27B release, the biggest change is quality. The footprint remains 5.9 GB, but the gap to full precision has narrowed materially, with particularly strong gains in agentic coding, multimodal reasoning, and long-horizon tool use.
Ternary Bonsai 2 27B is available today under Apache 2.0.
Local AI shouldn't be limited to expensive hardware. Everyone should have the opportunity to own their AI.
tpae retweeted
Raptor 0.6 just shipped.
Most Macs have 8 or 16 GB. Most AI models pretend they don't. Raptor is our agent model built for those machines. Reads folders, checks spreadsheets, drafts emails. Local, free.
3.7 GB, down 41% from the last version.
tpae retweeted
Windows users deserve a great AI app that doesn't hog their system resources. I built the @Windows version of Osaurus to be fully native with a focus on user experience, performance, and ease of use with local models.
I can't wait for you to try it and share your feedback!
First fully native native agent harness for @Windows. Feel the difference with native software.
tpae retweeted
It's here. The first fully native agent harness, built for Windows. No electron in sight.
Osaurus for Windows is almost ready for beta. Get ready for a team of agents running right on your PC, local and private.
Watch Qwen3.8-27B set up a team of agents to perform a code review.
tpae retweeted
We've been building exactly this while being fully open source and MIT licensed.
In Osaurus, you can build a team of agents where each one has its own memory, tools, and skills. An orchestrator can delegate work to them as sandboxed subagents. You can tie them together with Projects to share the same instructions, knowledge base, and memory across every chat.
Our harness is fully model agnostic (local or bring your own cloud key)
github.com/osaurus-ai/osauru…
I will be in SF today and tomorrow! Looking to meet up and talk about local AI 😊
tpae retweeted
Its time we focus on sub 24gb Mac’s - in this day and age, when someone goes and buys a $3000 Mac, while us local-pilled people understand that nearly all that matters is RAM and its speed, the main majority of people in the world consider and expect it to be a high end device - which when compared to the amount of raw compute you’d be able to get 5-10 years ago for that cost, it is.
I will be focusing on our “mixture of models” theory; where a main orchestration/chat model spawns and delegates specifics tasks over to another model which specializes in that task and proceeds to act in a manner of;
Main model 1 is unloaded -> task specific model 2 is loaded -> does task -> unload model 2 -> load model 1 back and resume
I hope that over time this will be able to prove that its not a high parameter count that matters, but that having a handful of models which hyperfocus on specific topics being loaded and unloaded in an efficient manner can come to meet the needs of all automation.
Super proud of the team for this one.
The hardest messages to read were always from people on base Airs and minis who wanted to run agents locally and couldn't. Not because they had bad machines. Because nobody was building models for them.
Now somebody is.