ffs guys GPT-5 is not AGI it is not even close to being AGI it is not going to become AGI it has the same flaws as all LLMs do but it is a great model possibly the best of today and, as always, this was just day 1 and I could be wrong

yep no model in the world comes anywhere close to this going sleep 100% certain I'm right on my judgement as always, this will soon be common sense, but I said it first (: see you

GPT-5 basically one-shot this

Nah you're all wrong, GPT-5 is a leap I'm 100% doubling down here I didn't want to post too fast and regret it again, but it just solved a bunch of very, very hard debugging prompts that were previously unsolved (by AI), and then designed a gorgeous pixelated Gameboy game with a level of detail and quality that is clearly beyond anything else I've ever seen. There is no way this model is bad. I think you're all traumatized of benchmaxxers, and over-compensating against a model that is actually good. I also think you're underestimating gpt-oss's strengths (but yeah my last post was rushed) I still don't know if it is usable for serious programming though (o3 wasn't), but it seems so? A coding model as reliable as Opus, yet smarter than o3, would completely change my workflow. Opus doesn't need thinking to be great though, so, that might weight in its favor. For what it is worth, I only really used 3 models: - Opus 4.1 for coding - Gemini 2.5 very rarely for coding when Opus fails - o3 for everything but coding

Nah you're all wrong, GPT-5 is a leap I'm 100% doubling down here I didn't want to post too fast and regret it again, but it just solved a bunch of very, very hard debugging prompts that were previously unsolved (by AI), and then designed a gorgeous pixelated Gameboy game with a level of detail and quality that is clearly beyond anything else I've ever seen. There is no way this model is bad. I think you're all traumatized of benchmaxxers, and over-compensating against a model that is actually really good. I also think you're underestimating gpt-oss's strengths (but yeah my last post was rushed) I still don't know if it is usable for serious programming though (4o, o3 definitely weren't), but it seems so? A coding model as reliable as Opus, yet smarter than o3, would completely change my workflow. Opus doesn't need thinking to be great though, so, that might weight in its favor. For what it is worth, I only really used 3 models: - Opus 4.1 for coding - Gemini 2.5 very rarely for coding when Opus fails - o3 for everything but coding

"preventing death is highly unethical"

preventing death is highly unethical

So gpt-oss 120B can't produce correct german, yet nails complex Haskell bugs that even Opus failed to identify? How does that happen? I'm genuinely so confused by all of this

Promising to myself I'll never again try to implement fast program search on Haskell / GHC. It is just impossible HVM3 is, unfortunately, incredible

My initial impression on OpenAI's OSS model is aligned with what they advertised. It does feel closer to o3 than to other open models, except it is much faster and cheaper. Some providers offer it at 3000 tokens/s, which is insane. It is definitely smarter than Kimi K2, R1 and Qwen 3. I tested all models for a bit, and got very decisive results in favor of OpenAI-OSS-120b. Unfortunately, there is one thing these models can't do yet - my damn job. So, hope you guys have fun. I'll be back to debugging superposed λ-calculus evaluation 😭 see you