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

GPT-4.1 in the API

openai.com

Post reply on HN
Print view View original

441–450 of 513 posts

Re: GPT-4.1 in the API

#441

Post by miki123211 »

Most of the improvements in this model, basically everything except the longer context, image understanding and better pricing, are basically things that reinforcement learning (without human feedback) should be good at.

Getting better at code is something you can verify automatically, same for diff formats and custom response formats. Instruction following is also either automatically verifiable, or can be verified via LLM as a judge.

I strongly suspect that this model is a GPT-4.5 (or GPT-5???) distill, with the traditional pretrain -> SFT -> RLHF pipeline augmented with an RLVR stage, as described in Lambert et al[1], and a bunch of boring technical infrastructure improvements sprinkled on top.

[1] https://arxiv.org/abs/2411.15124

Re: GPT-4.1 in the API

#442

Post by jjani »

XCSme wrote: post #274

I tried 4.1-mini and 4.1-nano. The response are a lot faster, but for my use-case they seem to be a lot worse than 4o-mini(they fail to complete the task when 4o-mini could do it). Maybe I have to update my prompts...

That sounds incredibly disappointing given how high their benchmarks are, indicating they might be overtuned for those, similar to Llama4.

Re: GPT-4.1 in the API

#443

Post by artdigital »

shortcord wrote: post #340
resters wrote: post #325

Earlier quoted context omitted.

I use them as follows: o1-pro: anything important involving accuracy or reasoning. Does the best at accomplishing things correctly in one go even with lots of context. deepseek R1: anything where I want high quality non-academic prose or poetry. Hands down the best model for these. Also very solid for fast and interesting analytical takes. I love bouncing ideas around with R1 and Grok-3 bc of their fast responses and…

Gemini 2.5 Pro is quite good at code. Has become my go to for use in Cursor. Claude 3.7 needs to be restrained too much.

Same here, 2.5 Pro is very good at coding. But it’s also cocky and blames everything but itself for something not working. Eg “the linter must be wrong you should reinstall it”, “looks to be a problem with the Go compiler”, “this function HAS to exist, that’s weird that we’re getting an error”

And it often just stops like “ok this is still not working. You fix it and tell me when it’s done so I can continue”.

But for coding: Gemini Pro 2.5 > Sonnet 3.5 > Sonnet 3.7

Re: GPT-4.1 in the API

#444

Post by muzani »

pton_xd wrote: post #224
swyx wrote: post #203

don't miss that OAI also published a prompting guide WITH RECEIPTS for GPT 4.1 specifically for those building agents... with a new recommendation for: - telling the model to be persistent (+20%) - dont self-inject/parse toolcalls (+2%) - prompted planning (+4%) - JSON BAD - use XML or arxiv 2406.13121 (GDM format) - put instructions + user query at TOP -and- BOTTOM - bottom-only is VERY BAD - no evidence that ALL CA…

As an aside, one of the worst aspects of the rise of LLMs, for me, has been the wholesale replacement of engineering with trial-and-error hand-waving. Try this, or maybe that, and maybe you'll see a +5% improvement. Why? Who knows. It's just not how I like to work.

One of the major advantages and disadvantages of LLMs is they act a bit more like humans. I feel like most "prompt advice" out there is very similar to how you would teach a person as well. Teachers and parents have some advantages here.

Re: GPT-4.1 in the API

#445

Post by taurath »

namaria wrote: post #433
qingcharles wrote: post #401

Earlier quoted context omitted.

Wow, I wondered what the limit was. I never checked, but I've been using it hesitantly since I burn up OpenAI's limit as soon as it resets. Thanks for the clarity. I'm all-in on Deep Research. It can conduct research on niche historical topics that have no central articles in minutes, which typically were taking me days or weeks to delve into.

I like Deep Research but as a historian I have to tell you. I've used it for history themes to calibrated my expectations and it is a nice tool but... It can easily brush over nuanced discussions and just return folk wisdom from blogs. What I love most about history is it has lots of irreducible complexity and poring over the literature, both primary and secondary sources, is often the only way to develop an understa…

LLMs seem fantastic at generalizing broad thought and is not great at outliers. It sort of smooths over the knowledge curve confidently, which is a bit like in psychology where only CBT therapy is accepted, even if there are many much more highly effectual methodologies on individuals, just not at the population level.

Re: GPT-4.1 in the API

#446

Post by muzani »

The real news for me is GPT 4.5 being deprecated and the creativity is being brought to "future models" and not 4.1. 4.5 was okay in many ways but it was absolutely a genius in production for creative writing. 4o writes like a skilled human, but 4.5 can actually write a 10 minute scene that gives me goosebumps. I think it's the context window that allows for it to actually build up scenes to hammer it down much later.

Re: GPT-4.1 in the API

#447

Post by miroljub »

SweetSoftPillow wrote: post #423
lxgr wrote: post #317

As a ChatGPT user, I'm weirdly happy that it's not available there yet. I already have to make a conscious choice between - 4o (can search the web, use Canvas, evaluate Python server-side, generate images, but has no chain of thought) - o3-mini (web search, CoT, canvas, but no image generation) - o1 (CoT, maybe better than o3, but no canvas or web search and also no images) - Deep Research (very powerful, but I have…

Switch to Gemini 2.5 Pro, and be happy. It's better in every aspect.

Warning to potential users: it's Google.

Re: GPT-4.1 in the API

#448

Post by simonw »

krat0sprakhar wrote: post #410
simonw wrote: post #182

Here's a summary of this Hacker News thread created by GPT-4.1 (the full sized model) when the conversation hit 164 comments: https://gist.github.com/simonw/93b2a67a54667ac46a247e7c5a2fe... I think it did very well - it's clearly good at instruction following. Total token cost: 11,758 input, 2,743 output = 4.546 cents. Same experiment run with GPT-4.1 mini: https://gist.github.com/simonw/325e6e5e63d449cc5394e92b8f2a3…

Hey Simon, I love how you generates these summaries and share them on every model release. Do you have a quick script that allows you to do that? Would love to take a look if possible :)

I use this one: https://til.simonwillison.net/llms/claude-hacker-news-themes
Post reply on HN