24, sde @ stealth startup, performative nerd

hyderabad
cheaty retweeted
Legend of Zelda: Ocarina of Time running natively on Jailbroken 9.00 PS5. Thank you @Phi10w for the idea.
8
13
115
4,708
nobody was gonna tell me one single code review with GPT-6 Astra is gonna cost me $10.18 on the API??? jesus christ man this is ONE PROMPT i should've just bought a chatgpt plus sub
27
2
201
8,399
someone at OpenAI should correct the title of this update on the OpenAI API platform GPT-5.6 Astra lol
3
1
108
4,414
if you buy a Claude Pro/Max sub today, you can still get a free reset :)
49
6
539
34,034
if you create a new account today, this will not work, as long as you have an existing account that either never had a subscription or it expired, you can still get your reset
1
13
1,665
wow you guys really hate these things that do all your work huh believing you can be disrespected by a computer is a certainly a choice
U gave it the option to disregard you by asking a question. You should’ve told it to stop and hand it off.
1
9
1,419
whatsapp muse
13
976
in an increasingly volatile world where LLMs get more general, i think it's basically a necessity to become a generalist in all areas of your life, an engineering background helps of course but that will not always be the case taste will matter for a while, but not for long
2
1
31
1,403
Australia has been hacked. 'And today, I spoke with the CEO of OpenAI, Sam Altman, to express Australia's extreme concern about this incident. And I also expressed my disappointment that it took the company way too long to inform the government what had occurred, and the nature of the way that that notification occurred as well was unacceptable.'
5
18
537
26,422
yikes, not looking good for OpenAI
Is GPT-6 Luna worse "vision" model than GPT-5.6 Luna? Yep. - extraction (high): ↓ 81.79% → 66.67% - counting (high): ↓ 70.72% → 64.41% - reasoning (high): ↓ 65.56% → 60.71% - detection (high): ↑ 62.29% → 64.12% link: playground.roboflow.com/shar… ↓ more examples
1
41
2,073
opus 5.5 is one cocky mf so far it has: >completely ignored me asking it several questions because it is hyperfocused on a bug fix it was failing to solve >i told it to give it to codex to figure out while it answers my previous query >ignores me again >fixes the bug
20
3
257
12,870
cheaty retweeted
Claude Opus 5.5 is out on VoxelBench not much to say except that it's wonderful GPT-6 Sol is also out, enjoy!
Claude Opus 5.5 has ranked 1st on VoxelBench and GPT-6 Sol ranks 3rd, improving over 5.6 competition is fierce among the top 3 models!
4
8
136
7,702
GPT-6 Sol (max) loses to Claude Opus 5 (max) (not Opus 5.5) on WebDev Arena 😬
8
1
71
3,205
New stealth model 'Space Bunny' is most certainly MiniMax M3.1. I can independently confirm that the text tokenizer perfectly matches that of the MiniMax M3. The image tokenizer seems different, likely upgraded.
A new Stealth model is being tested on OpenRouter and OpenCode: "Space-Bunny-Alpha". When asked in Chinese, the model claims to be a MiniMax model. Might be MiniMax M3.1. I haven’t done thorough testing yet, though, so take that with a grain of salt
22
8
375
21,886
more mentions of M3.1:
MiniMax M3.1 spotted in the official MiniMax Model Catalog. ty to @CaryPalmerr for pointing me in the right direction while we were digging around
8
1,091
MiniMax M3.1 spotted in the official MiniMax Model Catalog. ty to @CaryPalmerr for pointing me in the right direction while we were digging around
New stealth model 'Space Bunny' is most certainly MiniMax M3.1. I can independently confirm that the text tokenizer perfectly matches that of the MiniMax M3. The image tokenizer seems different, likely upgraded.
3
3
79
5,061
i just had Opus 5.5 get prompt injected by the safety classifier mid thinking and it rejected my request to build a very basic bulk tokenizer tester? goddamn it Anthropic, don't do this, this is not the way to do the classifier stuff, either the model thinks it's okay and does it or it doesn't do it :/
5
5
57
3,270
lmao this is all it takes to bypass said classifier
9
405
Gemini 3.8 Flash and Flash Lite TTS is launching imminently.
2
15
1,204
Vision tokenizer is a perfect match too, this is most certainly MiniMax-M3.1!
30
1,593