Sabitlenmiş Tweet
Dylan
1.4K posts
Dylan
@DylanNeve10
4th Year Computer Engineering @tcddublin && AI Software Engineer Intern @intel ..... all views are my own :)
Katılım Eylül 2022
355 Takip Edilen268 Takipçiler
@MishaalRahman Please tell me the widgets are fixed 😭 it's been months
English
2️⃣ The second beta of Android 17 QPR1 is here!
Android 17 QPR1 Beta 2 (CP31.260423.012.A1) is rolling out to devices enrolled in the Android Beta program. Check those updates!
English
@noahzweben @garrytan The community and the open source collaboration and hundreds of third party tools and integrations? Cowork will never be OpenClaw
English
@garrytan How can we make cowork more powerful? What feels most missing
English
Claude Cowork is powerful but right now nothing compared to OpenClaw
It’s a race. May the best product win.
Hosea@hoseakidane_
@garrytan Gary, have you tried claude cowork for the same tasks you would want openclaw for? A lot of the features overlap now like using iMessage, scheduling, using computer use etc
English
@trq212 This is horrible I'm constantly hitting the rate limit on 5x Max now in under half the 5h session when I never had a single issue before. The $100 plan has become almost useless. I won't be able to justify renewing it if the current limits stay in place. Please reconsider
English
To manage growing demand for Claude we're adjusting our 5 hour session limits for free/Pro/Max subs during peak hours. Your weekly limits remain unchanged.
During weekdays between 5am–11am PT / 1pm–7pm GMT, you'll move through your 5-hour session limits faster than before.
English
so 3x the training compute gets you 1% improvement on swe bench multilingual and 21% on terminal bench 2.0 but k2.5 is in non thinking mode? if those benchmarks are useless, it's weird that they are the ones reported in cursor blog then?
something is wrong
Lee Robinson@leerob
Yep, Composer 2 started from an open-source base! We will do full pretraining in the future. Only ~1/4 of the compute spent on the final model came from the base, the rest is from our training. This is why evals are very different. And yes, we are following the license through our inference partner terms.
English
Yep, Composer 2 started from an open-source base! We will do full pretraining in the future.
Only ~1/4 of the compute spent on the final model came from the base, the rest is from our training. This is why evals are very different.
And yes, we are following the license through our inference partner terms.
Fynn@fynnso
was messing with the OpenAI base URL in Cursor and caught this accounts/anysphere/models/kimi-k2p5-rl-0317-s515-fast so composer 2 is just Kimi K2.5 with RL at least rename the model ID
English
the Codex app now has themes
choose from some classics like Catppucin, Monokai, and Solarized. or try themes built with some partners from Linear, Notion, and OpenClaw among others. or make your own and share it
this was a bit of a personal side quest that i’ve been thinking about for a long time. we’re doing themes a bit differently! a few notes:
English
Some Pixel Watch users say March update broke SpO2 and skin temperature tracking
✅ Details - piunikaweb.com/2026/03/09/pix…
English
Dylan retweetledi
Claude was the fastest-growing Gen AI tool by website visits in February.
English
@Enscion25 Can't even give me an answer though??? While others are calling out openai for the pain they are inflicting on people and abusing their own customers. You sit by and watch it happen, continuing to line their pockets. Why are you still giving them money????
English
@Enscion25 Keep supporting a psychopath and funding their endeavors so you can hurt more people in the community you claim to ""care about""
Disgusting
English
@synthwavedd GPT is great for people who can't code and have no understanding of what they're doing. As a collaborative partner for a professional developer Claude is the only option. It's pretty obvious imo
English
Working with GPT-5.4 in the API?
We’ve updated our prompting guide with patterns for reliable agents covering tool use, structured outputs, verification loops, and long-running workflows.
developers.openai.com/api/docs/guide…
English
GPT-5.4 is a big step forward
Eric Hartford@QuixiAI
GPT-5.4 is really good. I immediately notice its boost in understanding and ability to solve problems quickly and completely. Using it to create a compiler, Claude Code is pretty much stumped. GPT-5.3 was making slow progress. But GPT-5.4 just *gets* it.
English
