Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

OpenAI O3-Mini

openai.com

Post reply on HN
Print view View original

591–600 of 944 posts

Re: OpenAI O3-Mini

#591

Post by twilightfringe »

chrismsimpson wrote: post #589
mvkel wrote: post #510

I've been using cursor since it launched, sticking almost exclusively to claude-3.5-sonnet because it is incredibly consistent, and rarely loses the plot. As subsequent models have been released, most of which claim to be better at coding, I've switched cursor to it to give them a try. o1, o1-pro, deepseek-r1, and the now o3-mini. All of these models suffer from the exact same "adhd." As an example, in a NextJS app,…

I’ve coded in many languages over the years but reasonably new to the TS/JS/Next world. I’ve found if you give your prompts a kind long form “stream of consciousness”, where you outline snippets of code in markdown along with contextual notes and then summarise/outline at the end what you actually wish to achieve, you can get great results. Think a long form, single page “documentation” type prompts that alternate be…

ha! good to confirm! I tend to do this, just kind of as a double-check thing, but never sure if it actually worked or if it was a placebo, lol.

Or end with "from the user's perspective: all the "B" elements should light up in excitement when you click "C""

Re: OpenAI O3-Mini

#592

Post by macNchz »

cruffle_duffle wrote: post #568
macNchz wrote: post #533

Earlier quoted context omitted.

You can start it with something like: aider --architect --model o1 --editor-model sonnet Then you'll be in "architect" mode, which first prompts o1 to design the solution, then you can accept it and allow sonnet to actually create the diffs. Most of the time your way works well—I use sonnet alone 90% of the time, but the architect mode is really great at getting it unstuck when it can't seem to implement what I want…

I really want to see how apps created this way scale to large codebases. I’m very skeptical they don’t turn into spaghetti messes. Coding is basically just about the most precise way to encapsulate a problem as a solution possible. Taking a loose English description and expanding it into piles of code is always going to be pretty leaky no matter how much these models spit out working code. In my experience you have t…

I see what you’re saying and I think that terming this “architect” mode has an implication that it’s more capable than it really is, but ultimately this two model pairing is mostly about combining disparate abilities to separate the “thinking” from the diff generation. It’s very effective in producing better results for a single prompt, but it’s not especially helpful for “architecting” a large scale app.

That said, in the hands of someone who is competent at assembling a large app, I think these tools can be incredibly powerful. I have a business helping companies figure out how/if to leverage AI and have built a bunch of different production LLM-backed applications using LLMs to write the code over the past year, and my impression is that there is very much something there. Taking it step by step, file by file, like you might if you wrote the code yourself, describing your concept of the abstractions, having a few files describing the overall architecture that you can add to the chat as needed—little details make a big difference in the results.

Re: OpenAI O3-Mini

#594

Post by brookst »

apparent wrote: post #322
explain wrote: post #316

Earlier quoted context omitted.

You have to pick one to continue the chat.

Why not always pick the one on the left, for example? I understand wanting to speed through and not spend time doing labor for OpenAI, but it seems counter-productive to spend any time feeding it false information.

My assumption is they measure the quality of user feedback, either on a per user basis or in an aggregate. I want them to interrupt me less, so I want them to either decide I’m a bad teacher or that users in general are bad teachers.

Re: OpenAI O3-Mini

#595

Post by dagelf »

Bjorkbat wrote: post #70

I have to admit I'm kind of surprised by the SWE-bench results. At the highest level of performance o3-mini's CodeForces score is, well, high. I've honestly never really sat down to understand how elo works, all I know is that it scored better than o1, which allegedly as better than ~90% of all competitors on CodeForces. So, you know, o3-mini is pretty good at CodeForces. But it's SWE-bench scores aren't meaningfully…

I think the innovation here is probably that its a much smaller and so cheaper model to run.

Re: OpenAI O3-Mini

#596

Post by prompt_overflow »

Plot twist:

1. they are trying to obfuscate deepscrape success

2. they are trying to confuse you. the benchmark margins are minimal (and meaningless)

3. they are trying to get time (with investors) releasing nothing-special-models in a predicted schedule (jan -> o3, feb -> o3-pro-max, march -> o7-ultra, and in 2026 -> OMG! we've reached singularity! (after spending $500B))

-

And at the end of the day, nothing changes for me and neither for you. enjoy your time out of this sickness ai hype. bruh!

Re: OpenAI O3-Mini

#597

Post by lumost »

usef- wrote: post #344
lumost wrote: post #206

Earlier quoted context omitted.

Capex was the theoretical moat, same as TSMC and similar businesses. DeepSeek poked a hole in this theory. OpenAI will need to deliver massive improvements to justify a 1 billion dollar training cost relative to 5 million dollars.

I don't know if you are, but a lot of people are still comparing one Deepseek training run to the entire costs of OpenAI. The deepseek paper states that the $5mil number doesn't include development costs, only the final training run. And it doesn't include the estimated $1.4billion cost of the infrastructure/chips Deepseek owns. Most of OpenAI's billion dollar costs is in inference, not training. It takes a lot of co…

Anthropic’s ceo was predicting billion dollar training runs for 2025. Current training runs were likely in the tens/hundreds of millions of dollars USD.

Inference capex costs are not a defensive moat as I can rent gpus and sell inference with linear scaling costs. A hypothetical 10 billion dollar training run on proprietary data was a massive moat.

https://www.itpro.com/technology/artificial-intelligence/dol...

Re: OpenAI O3-Mini

#598

Post by dagelf »

futureshock wrote: post #511
cjbarber wrote: post #14

The interesting question to me is how far these reasoning models can be scaled. With another 12 months of compute scaling (for synthetic data generation and RL) how good will these models be at coding? I talked with Finbarr Timbers (ex-DeepMind) yesterday about this and his take is that we'll hit diminishing returns – not because we can't make models more powerful, but because we're approaching diminishing returns in…

I think in a lot of ways we are already there. Users are clearly already having difficulty seeing which model is better or if new models are improving over old models. People go back to the same gotcha questions and get different answers based on the random seed. Even the benchmarks are getting very saturated. These models already do an excellent job with your homework, your corporate PowerPoints and your idle questi…

Don't underestimate how much the long tail means to the general public.

Re: OpenAI O3-Mini

#599

Post by danparsonson »

throw83288 wrote: post #526
vok wrote: post #404

Well, o3-mini-high just successfully found the root cause of a seg fault that o1 missed: mistakenly using _mm512_store_si512 for an unaligned store that should have been _mm512_storeu_si512.

How do I avoid the angst about this stuff as a student in computer science? I love this field but frankly I've been at a loss since the rapid development of these models.

For all the value that they bring, there is still a good dose of parlour tricks and toy examples around, and they need an intelligent guiding hand to get the best out of them. As a meat brain, you can bring big picture design skills that the bots don't have, keeping them on track to deliver a coherent codebase, and fixing the inevitable hallucinations. Think of it like having a team of optimistic code monkeys with terrible memory, and you as the coordinator. I would focus on building skills in things like software design/architecture, requirements gathering (what do people want and how do you design software to deliver it?), in-depth hardware knowledge (how to get the best out of your platform), good API design, debugging, etc. Leave the CRUD to the robots and be the brain.

Re: OpenAI O3-Mini

#600

Post by 4b11b4 »

delichon wrote: post #577
torben-friis wrote: post #527

Earlier quoted context omitted.

My experience with cursor and sonnet is that it is relatively good at first tries, but completely misses the plot during corrections. "My attempt at solving the problem contains a test that fails? No problem, let me mock the function I'm testing, so that, rather than actually run, it returns the expected value!" It keeps doing that kind of shenanigans, applying modifications that solve the newly appearing problem whi…

Claude makes a lot of crappy change suggestions, but when you ask "is that a good suggestion?" it's pretty good at judging when it isn't. So that's become standard operating procedure for me. It's difficult to avoid Claude's strong bias for being agreeable. It needs more HAL 9000.

I'm always asking Claude to propose a variety of suggestions for the problem at hand and their trade-offs, then evaluating them for the top three proposals and why. Then I'll pick one of them and further vet the idea
Post reply on HN