chanatun | building @ 16 (@chanatundev) profile picture
Joined March 2026
Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)Photo by chanatun | building @ 16 (@chanatundev)
yet another model unavailable to students, thanks
Gemini 4 Hydrogen is here, bringing cutting-edge intelligence to everyone for fast, everyday help. It makes it easier to get things done, following your instructions more closely, calling the right tools, and getting you to a reliable output with less back-and-forth. 🧵

43
Grok Bot v1.0 is out of beta, and it seems to be moving toward collaborative bots (maybe enterprise): for us unenterprised X-lings: > more proactive and spots unfinished work, offers to take it > fewer needless approval cards > faster+more reliable browser, docs, and slides; better at interactive pages mainly enterprise: > you can now share your bots (multi-person threads) > integrates w/ Slack + can now join Google Meet, Zoom, and Teams; can chat in the call and send a summary > shared app logins x.ai/news/grok-bot-1-0
2
1
100
misinformation. its like comparing apple's dictation to wisprflow/willow voice. @AarushYadav47 literally talks about this in his post: teapaw.org/AarushYadav47/status/2… "IMUs are very limited. We tried IMUs on glasses, masks, and earbuds. However, IMUs measure your aggregated facial movement. It’s like being asked to develop Face ID but you’re only allowed a 1px by 1px image. It works for a few discrete classes like classifying between 5 words, because your head moves differently during yes and no, but carry that over to classifying 10K words or even on a phoneme basis, and it breaks down. The precision issue is fatal. We quickly realized it's not the path to continuous speech decoding.
startup killed before it actually started
2
135
This is a really good idea. ADEs are currently shipped as-is and adding features requires cloning and building. I hate restarting ADEs mid-thread, and this bsaically solves it for me. Thanks
Your coding agent can build anything except the app you use it in So I made Seed It starts as one chat. Ask for a feature and it builds it into itself. Works with your Claude Code, Codex, Cursor or OpenCode login and it's open source 👀
1
1
3
89
Google won?? > New SOTA Google Model > Best at cybersecurity > Best at long context > Best knowledge work and general work Unless its benchmaxxed like Argon, because if they do that, I think our trust is Google becomes unrepairable.
Gemini 4 Carbon is our next frontier model. It’s state-of-the-art across cybersecurity and long-horizon capabilities.
26
I find it funy how the News section for Anthropic's Claude Duet is filled with @bot posts and comments. Is XAI/XSI leveraging X.com to keept @bot no.1 lmao
1
44
pic from teapaw.org/xoaanya/status/2111944… not a good deal compared to the rest of the frontier
Google AI Ultra is such a good deal because how good Argon is! Will give out Ultra to folks who build the craziest things with Argon.
1
25
What happened to @datacurve's deepSWE, my favorite bench? They used to be so on top of things: > they even got GPT-6 Astra benches out on the same day it was released, It's been over a month now and nothing has changed at deepswe.datacurve.ai.
2
1
43
To be fair to the team, Gemini 3.8 seems to have bench-maxed DeepSWE. Hopefully they are cooking up a better bench because of this.
1
19
gemini, don't use google search! > uses duck duck go search > proceeds to use google search we're going to look back at gemini 3.1 pro and laugh oh wait we already do
3
94
this aged badly. gemini 4 pro just absolutely destroyed every frontier model. google is actually on the frontier again after years of humiliation
and then gemini 4.0 pro launches and gets mogged by qwen 27b 😭
1
1
99
oh my openai actually has a 'one more thing' prepared for the keynote
2
54
yeah openai started recursive self imporvement already its over
1
45
openai is ALREADY killing jev with their decisions api
46
theres almost no drop in quality in motion graphics from opus 5.5 to sonnet 5.5 this is insane
2
1
2
239
prompt: make a dynamic 15-second motion graphics video that shows what an incredible motion designer you are, like it's your showreel for a résumé. go all out. sonnet 5.5 didn't realize that it was supposed to be for itself lol
1
67
tomorrow is fable 5.5 during openai devday, right?
Introducing Claude Sonnet 5.5, the second model in the Claude 5.5 family. It’s a clear upgrade over Sonnet 5, runs more than 30% faster, and costs up to 30% less for most work.
35
might have to eat my words
and then gemini 4.0 pro launches and gets mogged by qwen 27b 😭
1
25