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

Building more with GPT-5.1-Codex-Max

openai.com

Post reply on HN
Print view View original

81–90 of 332 posts

Re: Building more with GPT-5.1-Codex-Max

#81

Post by awad »

nico wrote: post #65
johnfn wrote: post #56

I've been using a lot of Claude and Codex recently. One huge difference I notice between Codex and Claude code is that, while Claude basically disregards your instructions (CLAUDE.md) entirely, Codex is extremely, painfully, doggedly persistent in following every last character of them - to the point that i've seen it work for 30 minutes to convolute some solution that was only convoluted because of some sentence I t…

> Claude basically disregards your instructions (CLAUDE.md) entirely A friend of mine tells Claude to always address him as “Mr Tinkleberry”, he says he can tell when Claude is not paying attention to the instructions on CLAUDE.md when Claude stops calling him “Mr Tinkleberry” consistently

Highly recommend adding some kind of canary like this in all LLM project instructions. I prefer my instructions to say 'always start output with an (uniquely decided by you) emoji' as it's easier to visually scan for one when reading a wall of LLM output, and use a different emoji per project because what's life without a little whim?

Re: Building more with GPT-5.1-Codex-Max

#82

Post by mrguyorama »

levocardia wrote: post #61
syntaxing wrote: post #46

I rarely used Codex compared to Claude because it was extremely slow in GitHub copilot . Like maybe 2-5X slower than Claude Sonnet. I really wish they just made their models faster than “better”

Very interesting to see the range of peoples' preferences. I would almost always prefer smart over fast; I have all my LLMs to be all-thinking-all-the-time.

GPT-5 was recently updated to make it more "thinking" and "warmer" or whatever and now a task (semantically compare these two short files) that used to take 5 seconds and reliably produce useful and consistent output now takes 90 seconds to "think" (while it's thinking output makes it pretty clear there is zero thinking happening) and produces a completely differently structured output every single time, making the tool not only slower and more expensive to use, but worse at a simple task that LLMs should be very good at.

There's an option to "get a quick answer" and I hoped clicking that would revert to previous performance and instead what it does is ignore that I uploaded two files and asks me to upload the files

Literally the only real good task I've found for these dumb things and they still found a way to fuck it up because they need to keep the weirdos and whales addicted. It's now almost easier to go back to comparing these files by eye, or just bite the bullet and finally write a few lines of python to actually do it right and reliably.

Re: Building more with GPT-5.1-Codex-Max

#83

Post by hadlock »

johnfn wrote: post #56

I've been using a lot of Claude and Codex recently. One huge difference I notice between Codex and Claude code is that, while Claude basically disregards your instructions (CLAUDE.md) entirely, Codex is extremely, painfully, doggedly persistent in following every last character of them - to the point that i've seen it work for 30 minutes to convolute some solution that was only convoluted because of some sentence I t…

I've been really impressed with codex so far. I have been working on a flight simulator hobby project for the last 6 months and finally came to the conclusion that I need to switch from floating origin, which my physics engine assumes with the coordinate system it uses, to a true ECEF coordinate system (what underpins GPS). This involved a major rewrite of the coordinate system, the physics engine, even the graphics system and auxilary stuff like asset loading/unloading etc. that was dependent on local X,Y,Z. It even rewrote the PD autopilot to account for the changes in the coordinate system. I gave it about a paragraph of instructions with a couple of FYIs and... it just worked! No major graphical glitches except a single issue with some minor graphical jitter, which it fixed on the first try. In total took about 45 minutes but I was very impressed.

I was unconvinced it had actually, fully ripped out the floating origin logic, so I had it write up a summary and then used that as a high level guide to pick through the code and it had, as you said, followed the instructions to the letter. Hugely impressive. In march of 2023 OpenAI's products struggled to draw a floating wireframe cube.

Re: Building more with GPT-5.1-Codex-Max

#84

Post by halifaxbeard »

amluto wrote: post #40

I would love to see all the big players put 1% of the effort they put into model training into making the basic process of paying and signing in suck less. Claude: they barely have a signin system at all. Multiple account support doesn’t exist. The minimum seat count for business is nonsense. The data retention policies are weak. OpenAI: Make ZDR a thing you can use or buy without talking to sales, already. And for t…

At this point I’m not convinced that Gemini 3 Pro was post-trained on data Google had permission to use, going by the myriad of issues on the Gemini CLI tracker around Google AI/Google One/Google Cloud/Google Workspaces.

https://github.com/google-gemini/gemini-cli/issues/12121

It is far too easy to accidentally end up under the wrong privacy agreement, to the point of where some workplaces are banning use of the Gemini CLI!

Re: Building more with GPT-5.1-Codex-Max

#85

Post by tunesmith »

I've been dealing with Codex CLI for a while and I love it, but I'm wondering if my thinking is just limited. While I'm starting discussions and creating plan docs, I've never been able to ask it to do anything that takes it longer than 25 minutes or so. Usually far less. I'm having trouble imagining what I can ask it to do that would make it take hours - like, wouldn't that require putting together an absolutely massive planning doc that would take hours to put together anyway? I'd rather just move incrementally.

Re: Building more with GPT-5.1-Codex-Max

#86

Post by byefruit »

computerex wrote: post #52
amluto wrote: post #40

I would love to see all the big players put 1% of the effort they put into model training into making the basic process of paying and signing in suck less. Claude: they barely have a signin system at all. Multiple account support doesn’t exist. The minimum seat count for business is nonsense. The data retention policies are weak. OpenAI: Make ZDR a thing you can use or buy without talking to sales, already. And for t…

Couldn't agree more about the google product offerings. Vertex AI? AI Studio? Maker studio? Gemini? The documentation is fragmented with redundant offerings making it confusing to determine what is what. GCS billing is complicated to figure out vs OpenAI billing or anthropic. Sad part is Google does offer a ChatML/OpenAI compliant endpoint to do LLM calls and I believe they in an experiment also reduced friction in g…

I've just found myself using OpenRouter if we need Google models for a project, it's worth the extra 5% just not to have to deal with the utter disaster that is their product offering.

Re: Building more with GPT-5.1-Codex-Max

#87

Post by joshstrange »

atonse wrote: post #49
amluto wrote: post #40

I would love to see all the big players put 1% of the effort they put into model training into making the basic process of paying and signing in suck less. Claude: they barely have a signin system at all. Multiple account support doesn’t exist. The minimum seat count for business is nonsense. The data retention policies are weak. OpenAI: Make ZDR a thing you can use or buy without talking to sales, already. And for t…

Agree 1,000%. I just won’t even waste my time with the google stuff cuz I can’t figure out how to pay with it. And that’s a problem everywhere at google. Our google play account is suspended cuz I can’t verify the company. It won’t let me cuz it says I’m not the owner. I’ve always been the owner of my company. For 18 years. There is no one else. Once some error said make sure the owner email matches your profile in g…

The "Owner" accounts in Google Play and Apple's App Store are so freaking annoying. The only time they make sense is for solo-founders and even then I've had issues. Now expand it to working at a larger company and it's a joke, a bad one. Oh sure, I'll just get the CEO (or other higher-up) to login and accept new agreements, that will be easy. Even more fun when you tell a client (who logged in exactly 1 time to set up the account) that they need to use a generic email (not a personal one or an employee-specific one), the ignore your suggestion, and then they can't get back in because the person who set up the account left the company. It's a mess.

Also, re "Google Payments", I tried to transfer an app from my personal/solo Google Play account to a new business one I set up for my LLC and it was like pulling teeth. They wanted me to find some payment id from the original $20 purchase I made to get access to Google Play, something I did right around when they first launched and while I still have/use the same email, Google came out with approximately 1 googol different "payment solutions" in the interim and their engineers don't care about data migrations. Finally, after many support emails, they just transferred it without me giving that code which just shows how silly the whole thing was from the start.

Re: Building more with GPT-5.1-Codex-Max

#88

Post by timtimmy »

amluto wrote: post #40

I would love to see all the big players put 1% of the effort they put into model training into making the basic process of paying and signing in suck less. Claude: they barely have a signin system at all. Multiple account support doesn’t exist. The minimum seat count for business is nonsense. The data retention policies are weak. OpenAI: Make ZDR a thing you can use or buy without talking to sales, already. And for t…

Google keeps changing their privacy and “don’t train on my data/code” options. When gemini-cli launched, there was a clear toggle for “don’t train on my code.” That’s now gone; it just links to a generic privacy page for me. Maybe something with my account changed, I can't figure it out. Deep in the Cloud Gemini console, there’s another setting that might control training, but it’s not clear what products it actually covers.

Trying to pay for Gemini-3 is confusing. Maybe an AI Ultra personal subscription? I already pay for OpenAI and Anthropic’s pro/max plans and would happily pay Google too. But the only obvious option is a $250/month tier, and its documentation indicates Google can train on your code unless you find and enable the correct opt-out. If that opt-out exists in all the products, it’s not obvious where it lives or what products it applies to.

Workspace complicates it further. Google advertises that with business workspace accounts your data isn’t used for training. So, I was going to try Antigravity on our codebase. At this point I know I can't trust Google, so I read the ToS carefully. They train on your prompts and source code, and there doesn't appear to be a way to pay them and opt out right now. Be careful, paying for Google Workspace does not protect you, always read the ToS.

Be careful with AI-studio and your Google Workspace accounts. They train on your prompts unless you switch it to API mode.

The result is a lot of uncertainty. I genuinely have no idea how to pay Google for Gemini without risking my code being used for training. And if I do pay, I can’t tell whether they’ll train on my prompts anyway.

The marketing for their coding products does not clearly state when they do or do not train on your prompts and code.

I had to run deep research to understand the risks with using Gemini 3 for agentic work, and I still don't feel confident that I understand the risks. I might have said some incorrect things above, but I am just so confused. I feel like I have a I don't have a lot of trust. And honestly, this feels confusing and deceptive. One could easily confuse it as deliberate strategy to gather training data through ambiguity and dark patterns, it certainly looks like this could be Google's strategy to win the AI race. I assume this is just how it looks, and that they aren't being evil on purpose.

OpenAI in particular has my trust. They get it. They are carefully building the customer experience, they are product and customer driven from the top.

Re: Building more with GPT-5.1-Codex-Max

#89

Post by sethhochberg »

temp0826 wrote: post #79
hassleblad23 wrote: post #57

Earlier quoted context omitted.

Adding to this, Google's models can only be used with GCP while OpenAI's models can be used with Azure, Anthropic's models can be used with AWD Bedrock, in addition to their own platforms. I'd love to see the Gemini models being available by other providers :) or if they just build a simple prepaid wallet like OpenAI and Anthropic.

Didn't realize these stipulations for the models. Looking at devops-y job descriptions the last few months I noticed nearly everyone has some kind of Azure requirement now (which I've mostly avoided because I don't want to end up managing someone's AD), but is openai the actual reason for it?

We're just using Github Copilot as our primary entrypoint for all of the model families. Its the only way we can easily offer our devs some level of Claude, Gemini, and Codex all in one place.

Re: Building more with GPT-5.1-Codex-Max

#90

Post by GenerWork »

tunesmith wrote: post #85

I've been dealing with Codex CLI for a while and I love it, but I'm wondering if my thinking is just limited. While I'm starting discussions and creating plan docs, I've never been able to ask it to do anything that takes it longer than 25 minutes or so. Usually far less. I'm having trouble imagining what I can ask it to do that would make it take hours - like, wouldn't that require putting together an absolutely mas…

Perhaps they're combining an incredibly complex product that has a lot of interactive features, a big codebase, test creation, and maybe throwing some MCP stuff in there such as creating creating a ticket in Jira if a test fails?
Post reply on HN