Am I the only one NOT excited about the 3D capabilities of 6.2 sol? I have a few tasks where Astra's 3D stuff was useful. 99% of the time I just want it to write code. Make the next generation of models write code so good @mitchellh stops reading the code. Then you'll impress me.
1
115
NGL, T3 code app is literal miles better experience than codex and Claude code... I don't usually drop a review but whoever built it on the team absolutely cooked.
Nitpick, wish it had the boring sidebar layout. I really tried the inbox style but it's just worse for me.
55
That's exactly what T3 Code does for you automatically behind the scenes. I just switch models and it hands over gracefully to the new model and session.
1
1
38
With the latest models from both anthropic and openai, changing effort levels no longer loses the cache
I switch models only to preserve usage on my CC sub for an upcoming task. Usually end up regretting it when GPT 6.1 inevitably butchers all of opus' hard work.
I use T3 Code
2
7
167
Big fan, so take this as feedback.
Why at 0:13s why does the "Use a template" tab take almost 2s to load? Even switching back to the "Import a repository" tab that you started on takes a similar amount of time to load.
Wish it all felt snappy and instant.
1
2
169
Gotcha, and then excuse my ignorance here but I assume it has skills or whatever (not looked into how plugins work) on how to profile an apps performance and guidance on how to fix performance bottlenecks fine tuned by the margelo team's expertise around stuff like this?
11
Have you solved the issue of me needing to wait 1min on a loading screen every time I open the discord app while it downloads and installs updates?
1
199
The graphs are really helpful for humans. And the plugin looks really good. I'm just wondering why I even need to visually identify the spikes and dips. Surely AI is smart enough to do that for me.
1
1
35
Genuine question, why can't I just tell Codex "Profile the performance of the app when _ then iteratively improve it while maintaining the same functionality until we reach a stable 60fps. Profile before and after each change."
1
1
136
Dope, going to see if I can swap this out for my DIY solution and pair it with @tan_stack DB the same way.
Gives me realtime, sever controlled data, simple inertia controllers with client side cacheing so my dashboard pages and tabs load instantly.
1
96
Is this because high overthinks less and makes less stupid mistakes than max?
I just put everything on max because brain sees higher number on Evals, must be better.
But I guess that doesn't translate here because evals don't test for the annoying stuff that makes it worse?
170
It's been 2 days. Your good buddy. It do feel like if you're not going at 1000mph nowadays you're moving to slow.
27