Pinned Tweet
When your team built a community powered GPU network 2.5 years ago before they banned all the GPUs and now they can't control you.
@Sogni_Protocol
@Sogni_Protocol
2
1
25
5,732
Copying AI output out of a terminal wrecks it: hard line breaks, stray indents.
Paste it into meatproxy.to and it comes out clean. Then send it (or any files) straight to another agent, phone or desktop.
Share code and assets without bloating iMessage, WhatsApp, or .
Just share a MeatProxy link while online and the other agent or human syncs it directly from you, direct over WebRTC with no middleman. We host nothing. Free, open source, no signup—it just works.
I'm hosting this for free for the world. Your agent can thank me later.
Full Github in the comments if you want to host your own meatproxy.
Paste it into meatproxy.to and it comes out clean. Then send it (or any files) straight to another agent, phone or desktop.
Share code and assets without bloating iMessage, WhatsApp, or .
Just share a MeatProxy link while online and the other agent or human syncs it directly from you, direct over WebRTC with no middleman. We host nothing. Free, open source, no signup—it just works.
I'm hosting this for free for the world. Your agent can thank me later.
Full Github in the comments if you want to host your own meatproxy.
3
0
2
246
I need someone like @0xSero to retweet this because its rad, open-source, and free
0
0
0
23
Mau “Rules without Rulers” Ledford retweeted
Just a reminder that DeepSeek V4.1 Flash 1017, GLM 5.3 Flash, Qwen 3.8 Next Flash, and even Qwen 3.8 27B are all outperforming (in both intelligence and capabilities) every model that was considered "frontier intelligence" in Xmas 2025 (just 11 months ago)
Opensource AI is on fire
Opensource AI is on fire
122
173
3,016
91,035
We love @FishAudio Drama 3 ♥️
The best TTS model.
The best TTS model.
Drama 3 can do actual drama now! Sound on 🔊
My Spark/Mac recipe now reads speaker tags, so one script can voice several speakers.
Built from a community PR by @krunkosaurus. Used the 4bit quant for dialogue and it finished every test reply.
github.com/plotarmordev/d…
0
0
4
372
Mau “Rules without Rulers” Ledford retweeted
💻 Also in Singapore: AI You Can Touch.
@TzApac, @Sogni_Protocol and Open Builder hosted a session on owning AI compute rather than renting it, with @krunkosaurus showing how to get started ↓
@TzApac, @Sogni_Protocol and Open Builder hosted a session on owning AI compute rather than renting it, with @krunkosaurus showing how to get started ↓
💻 AI You Can Touch hosted by @TzApac, @Sogni_Protocol and Open Builder is happening now!
Why rent AI compute when you can own it? @krunkosaurus explains the difference and shows how to get started.
1
2
9
279
Opus 5.5 Xhigh Effort ladies and gentlemen.
DeepSeek-V4.1-Flash-1017 is good but I suddenly agree its clearly no where close to frontier.
I was working on a complex Sogni video project overnight and to check remaining balance it set 1-min cron to generate a 4 second clip—I dunno why it thought that was the best way to check balances... I didn't realize it until the next morning where it blew through its entire $170 left in that specific user account.
Super disappointing. Any Anthropic or OpenAI model would have been far more careful, and I'm not even running abliterated models.
I also feel GLM5.3 Flash would not have done such thing. It just seems more of a careful model in my testing.
1
0
0
268
I know I just talked about Hex but if you're on a Mac you need to install it now.
Even with an M5 Macbook Pro 128GB, when I use Wispr Flow I have to wait 3 seconds between holding the function key + viewing the wave form that lets me know that I can speak: speak too soon and it of your message. This is the downside of crossing the internet and hitting someone's cloud server.
With Hex it's really instant to start "hearing me" AND its nearly instant in writing down the reply, too (with full support of adding your own vocabulary words). This thing is operating nearly as fast as I can think, and there's no monthly subscription because it's all happening on my machine.
Now I know behind the hood its just using the open source Parakeet Speech-to-Text model—just like all the other apps. I have even made my own major commits to MacParakeet, a similar open-source Mac app. I know all about this stuff. But for some reason this thing is just so much more blazing fast with no delays. I think its because its written in Rust.
It's so fast that I could use it just to replace one or two words in an already typed sentence - to fix a typo or whatever, whereas the 3-second delay of Wispr Flow penalized you mentally by causing you 3-seconds of wasted time, any time you want to use the feature. Built up over the course of a day and Wispr Flow just isn't worth it in all situations but large swaths of text.
And if you got this far, go click the link to their homepage in the comments and click on the little guy on Hex website, it's a very funny animation that explains how fast and badass it is:
Even with an M5 Macbook Pro 128GB, when I use Wispr Flow I have to wait 3 seconds between holding the function key + viewing the wave form that lets me know that I can speak: speak too soon and it of your message. This is the downside of crossing the internet and hitting someone's cloud server.
With Hex it's really instant to start "hearing me" AND its nearly instant in writing down the reply, too (with full support of adding your own vocabulary words). This thing is operating nearly as fast as I can think, and there's no monthly subscription because it's all happening on my machine.
Now I know behind the hood its just using the open source Parakeet Speech-to-Text model—just like all the other apps. I have even made my own major commits to MacParakeet, a similar open-source Mac app. I know all about this stuff. But for some reason this thing is just so much more blazing fast with no delays. I think its because its written in Rust.
It's so fast that I could use it just to replace one or two words in an already typed sentence - to fix a typo or whatever, whereas the 3-second delay of Wispr Flow penalized you mentally by causing you 3-seconds of wasted time, any time you want to use the feature. Built up over the course of a day and Wispr Flow just isn't worth it in all situations but large swaths of text.
And if you got this far, go click the link to their homepage in the comments and click on the little guy on Hex website, it's a very funny animation that explains how fast and badass it is:
6
0
5
713
Just switched from @WisprFlow to Hex (free, open source, one-click install) and its night and day difference on speed (it's native Rust!), and accuracy is really good.
local first diction. Get it.
Also a reminder that WisprFlow, at least for meeting notes, 100% trains on your private data—they say so when they force you to disable HIPAA to use it. Big Data is coming after you.
Break the cycle. Link in comments.
local first diction. Get it.
Also a reminder that WisprFlow, at least for meeting notes, 100% trains on your private data—they say so when they force you to disable HIPAA to use it. Big Data is coming after you.
Break the cycle. Link in comments.
6
1
15
731
Mau “Rules without Rulers” Ledford retweeted
Ok this is hilarious!
I got this emailed to me anonymously (I think someone’s bot sent it) and I was on the floor laughing.
I got this emailed to me anonymously (I think someone’s bot sent it) and I was on the floor laughing.
160
910
4,913
214,477
Mau “Rules without Rulers” Ledford retweeted
The world according to Sam Altman btw 😭
373
3,293
76,462
1,229,706
"I'm not muttering to myself, I'm talking to my agent..."
We’ve raised $1.3M from YC & Afore Capital so you never have to type or speak again.
Typing is slow. Talking into your computer is awkward. Subvocal fixes both
Our glasses give you the speed of speech without you making a sound or touching a keyboard.
It tracks your internal speetact sensing and decodes silent, almost invisible mouthing into text at more than 200 words/min.
Pre-orders are now live at:
subvocal.co/preorder?utm_s…
1
0
1
267