Joined January 2017
Hopped on Reddit and see this AI vegan crying about AI being too good at decompiling games and that he's shutting down his site in protest. Redditors shrug and post vibe coded clones. We are witnessing history.
39
81
9
2,421
68,965
automatic prompts have NO chill. And just so no one thinks its tailored to your behavior, I never prompt like that.
11
Bitch I'm in America with the $200 plan. Where tf IS this available?
1
40
We're giving away a CONTROL Resonant custom wrapped NVIDIA GeForce RTX 5080 to celebrate its official release with #RTXON. To enter: 馃煝 Share this post 馃煝 Comment #RTXON T&Cs: nvda.ws/4z24Gcq
37,660
31,173
2,817
33,620
3,055,808
The amount of wow factor Opus 5.5 gives you is entirely inversely correlated to how much you have used GPT Astra
31
I have Astra and Opus 5.5 working together on a decompilation project. One thing I've noticed: Astra is way more precise and thorough. It is constantly correcting Opus, who will check, then agree.
30
Why did this take 1 minute to answer!? This should be in memory. I'm assuming the CoT was hundreds of back and forth "wait...I should answer Codex..but maybe...wait..."
19
Using Codex (or Claude) to instantly create small game mods or tweak existing ones is peak. I wonder how badly Nexus and other mod sites have been hit.
1
31
Also you can just copy errors into Codex and instantly get a fix for your specific bug. Remember when we had to wait weeks for bug fixes?
15
rempred retweeted
Do you want to be free?
31
16
6
208
82,710
Holy shit @OpenAI fixed over-engineering
15
I asked Astra to fix my N64 decomp repo. It spawned 6 top level Sol threads, coordinated review and implementation, had the 6 thread communicating with themselves instead of wasting the directors turn. All in Codex threads, not as sub agents, and in the same worktree. @ChatGPT
78
Astra/Codex could take over 90% of office job right now
22
Ox Alpha is clearly inferior to Sol. But visible CoT makes it almost more valuable to use.
53
Ox Alpha isn't as good as Sol. Maybe on par with Terra
52
I need to know why there are thousands of accounts posting the most obvious AI slop under every thread agreeing with the premise or making their dumb AI point. Why are they doing this? Do they think they're going to actually make money?
1
50
"This run is instrumentation-contaminated" 馃馃馃
35
This little "More Details" button uses 5.5 to explain stuff. The new models are so bad a writing they had to regress an explanation LLM into the harness.
38
Tell your agent your app has a bug (make one up) and watch it find and fix it.
24
Luna is pretty dumb compared to Sol but it does a good job when you give it very specific, checkable tasks with guardrails
23