Own your AI. Local-first agents that remember, run code safely, and reach you anywhere. Native on Mac and Windows. Any model. No cloud required. Open source.

Joined May 2025
This model is small, fast, and pretty smart too.
Journalists can't upload a leak to the cloud. So we turned off the Wi-Fi. A 1.7B model read the whole case file, followed the money, and cited its sources. LFM3. Day 0 on Osaurus.
2
3
376
Osaurus retweeted
We built the investigative journalist agent to test one thing: can a tiny model do real work on sensitive files? LFM3 did it on day 0, with the Wi-Fi off.
Journalists can't upload a leak to the cloud. So we turned off the Wi-Fi. A 1.7B model read the whole case file, followed the money, and cited its sources. LFM3. Day 0 on Osaurus.
1
5
393
Osaurus retweeted
speed is the name of the game with lfm models truly any mac can use.
Journalists can't upload a leak to the cloud. So we turned off the Wi-Fi. A 1.7B model read the whole case file, followed the money, and cited its sources. LFM3. Day 0 on Osaurus.
1
3
24
1,832
Watch my phone work with my PC to generate an image without ever using a cloud model. Chat: Qwen3.8-27B Image: FLUX.3 Image Klein
Get Osaurus on your phone now! Use your Mac/PC to do all the heavy lifting and inference, while on the go with your iPhone or ipad. Also, dinos.
1
2
6
422
Get Osaurus on your phone now! Use your Mac/PC to do all the heavy lifting and inference, while on the go with your iPhone or ipad. Also, dinos.
Call your agent. Osaurus is now on iPhone and iPad.
1
1
4
632
Osaurus retweeted
Can't wait to ship everything that we've planned for the iOS app! Stay tuned :)
Call your agent. Osaurus is now on iPhone and iPad.
1
1
5
212
Osaurus retweeted
I call my agent more than I call my friends now. Huge work by @RaajeevCn on this one.
Call your agent. Osaurus is now on iPhone and iPad.
2
5
310
Osaurus retweeted
Nanbeige actually lived up to its expectations, with reasoning on it tended to over reason (not in a way of thinking about things that didn't matter too much like the Qwen series, but much more in a "plans out all the specifics even in a simple job" kind of way), and while I had it turned off for the demo, I know just from the tests I did that it lives up to the claims and competes with Qwen 3.8 27b as a heavy coding capable model. While cache may be larger due to older arch, Osaurus' efficient cache handling allows for smooth turn by turn usage without your RAM usage going higher than the model's ~20gb.
Nanbeige4.3-33B, day 1 on Osaurus. Neoclouds are dying. It made a game about it. On a Mac.
2
3
19
1,696
Osaurus retweeted
One prompt. It read the news, planned it in a todo list, wrote the game, and tested it before handing it over.
Nanbeige4.3-33B, day 1 on Osaurus. Neoclouds are dying. It made a game about it. On a Mac.
2
6
354
Osaurus retweeted
Excited for this. From what I can tell the scores alone show it rivals models like glm53flash. Osaurus users were having good results with Nanbeige 4b so hoping to see how this does even though I know the whole looped thing will make it dirty slow
Replying to @nanbeige
@nanbeige new 33B model beats qwen 3.8 27b Also, supports a CTX length up to ~524k! huggingface.co/Nanbeige/Nanb…
4
2
38
3,831
Osaurus retweeted
The scrolling performance in the chat has been carefully crafted to handle long conversations. You should be able to scroll without any hitches even when a response is actively streaming :) You should be able to feel the same performance on all iPhones (and iPads)!
1
1
3
259
Osaurus retweeted
zero latency is all you need. go local
yes. zero latency on my macbook
2
5
420
Osaurus retweeted
Windows users who are tired of electron bloatware should definitely try out what @SonofNun has come up with!
It’s now public! Now invite required. Try it out and tell me what you think! Pairs great with @NVIDIARTXSpark devices like the @surface Laptop Ultra!
There's a new version of this post
1
3
158
Osaurus retweeted
Should be a breeze
1
3
222
RT @SonofNun: It’s now public! Now invite required. Try it out and tell me what you think! Pairs great with @NVIDIARTXSpark devices like…
This quoted post is unavailable.
1
110
Osaurus retweeted
Privacy is a fundamental human right. Not because you have something to hide. Because you deserve a choice.
2
6
297
Osaurus retweeted
Here's a snippet of the @OsaurusAI iOS app that I'm currently working on. It's being built entirely in UIKit with no traces of SwiftUI for maximum performance and backward compatibility. Excited to bring the Mac app to mobile with Day 0 support for the new iPhone Duo!
5
2
2
14
1,949
Osaurus retweeted
Thanks for calling local models the frontier. Intelligence per watt is improving fast on its own curve. Soon local models will match what feels like frontier today. When that happens, there’s an enormous opportunity: tools, infrastructure, and support for people building businesses on the electrons they already paid for. Labs like @ornith_ , @Alibaba_Qwen , @liquidai , and @nanbeige are already pushing this forward with almost no revenue model. We’re building the infrastructure for it. Reach out if you want to learn more.
Today, a16z Crypto is becoming a16z Frontier. Since joining a16z in 2013, @cdixon has focused on technologies that look futuristic well before they reach the mainstream. Those themes led to investments in Oculus, Skydio, Zipline, Coinbase, Solana, and Uniswap. Frontier expands that mandate beyond crypto to back "outer planet" founders who are building new technologies and creating markets that may take years to develop. Chris joins @pmarca, @bhorowitz, and @eriktorenberg to discuss why crypto has become part of the broader tech landscape, why Frontier fills a hole in the firm's architecture, and why they'd rather be too early than too late. They also get into why the gap between AI-native startups and incumbents is the widest Marc has ever seen, how AI could pull forward 20 years of frontier technology, why stablecoins may become the way AI agents pay each other, how AI is breaking the internet's business model, and why today's 18-year-olds are about to become superhumans. 0:50 Introducing a16z Frontier 2:49 Smartphone parts to drones and VR 4:47 Backing the "outer planets" 7:11 Good ideas that look like bad ideas 10:43 The JP Morgan photo in Marc's office 15:13 Good tech ideas eventually work 16:53 AI may pull tech 20 years forward 18:18 The widest AI gap Marc's ever seen 22:06 Workers adopt AI before companies do 26:32 Old use cases vs. AI-native ones 30:08 Beyond "product manager brain" 34:45 AI can reverse engineer any hardware 38:23 The case for open source AI 46:36 How AI breaks the internet's model 49:11 VR, homebrew hacking, multiplayer AI 54:00 If you're 18, study "modern alchemy" YouTube: youtu.be/-9XQMHhpypc Not an offer or solicitation. None of the information herein should be taken as investment advice; Some of the companies mentioned are portfolio companies of a16z. Please see a16z.com/disclosures for more information. A list of investments made by a16z is available at a16z.com/portfolio.
1
1
8
1,071
Osaurus retweeted
The page isn’t the impressive part. Halfway through, the shell failed. It didn’t stop or ask for help. It switched to file tools and finished the job. 2.5B active params did that.
NVIDIA spent its keynote selling AI factories. A Mac read the whole thing and built a webpage about it. The model found itself in the coverage. Flint 40B. Running locally. Native Swift. No Electron.
1
3
263