Interesting, the new model's prompt is ~half the size (10KB vs. 23KB) of the previous prompt[0][1].
SWE-bench performance is similar to normal gpt-5, so it seems the main delta with `gpt-5-codex` is on code refactors (via internal refactor benchmark 33.9% -> 51.3%).
As someone who recently used Codex CLI (`gpt-5-high`) to do a relatively large refactor (multiple internal libs to dedicated packages), I kept running into bugs introduced when the model would delete a file and then rewrite in (missing crucial or important details). My approach would have been to just the copy the file over and then make package-specific changes, so maybe better tool calling is at play here.
Additionally, they claim the new model is more steerable (both with AGENTS.md and generally). In my experience, Codex CLI w/gpt-5 is already a lot more steerable than Claude Code, but any improvements are welcome!
So you're all saying suddenly codex cli w gpt 5 codex is better than claude code? Hard to believe
jumploops · 44m ago
Not suddenly, it's been better since GPT-5 launched.
Prompting is different, but in a good way.
With Claude Code, you can use less prompting, and Claude will get token happy and expand on your request. Great for greenfield/vibing, bad for iterating on existing projects.
With Codex CLI, GPT-5 seems to handle instructions much more precisely. It won't just go off on it's own and do a bunch of work, it will do what you ask.
I've found that being more specific up-front gets better results with GPT-5, whereas with Claude, being more specific doesn't necessarily stop the eagerness of it's output.
As with all LLMs, you can't compare apples to oranges, so to clarify, my experiences are primarily with Typescript and Rust codebases.
srcreigh · 34m ago
Codex CLI of course will sometimes do the wrong thing, or sometimes do something extra that you didn't intend for it to do.
It seems about half my sessions quickly become "why did you do that? rip __ out and just do ___". Then again, most of the other sessions involve Codex correctly inferring what I wanted without having to be so specific.
tedsanders · 3h ago
> SWE-bench performance is similar to normal gpt-5, so it seems the main delta with `gpt-5-codex` is on code refactors
SWE-bench is a great eval, but it's very narrow. Two models can have the same SWE-bench scores but very different user experiences.
Here's a nice thread on X about the things that SWE-bench doesn't measure:
Interestingly, "more steerable" can sometimes be a bad thing, as it will tend to follow your prompt to the letter even if that's against your interests. It requires better prompting and generally knowing what you're doing - might be worse for vibe-coders and better for experienced SWEs.
jumploops · 40m ago
Yes, given a similarly sparse prompt, Claude Code seems to perform "better" because it eagerly does things you don't necessarily know to ask
GPT-5 may underwhelm with the same sparse prompt, as it seems to do exactly what's asked, not more
You can still "fully vibe" with GPT-5, but the pattern works better in two steps:
1. Plan (iterate on high-level spec/PRD, split into actions)
2. Build (work through plans)
Splitting the context here is important, as any LLM will perform worse as the context gets more polluted.
htrp · 3h ago
think they're indexing here for professional work (people in the VSCode terminal)
klipklop · 16m ago
From my observation of the past 2 weeks is that Claude Code is getting dramatically worse and super low usage quota's while OpenAI Codex is getting great and has a very generous usage quota in comparison.
For people that have not tried it in say ~1 month, give Codex CLI a try.
king_magic · 57s ago
[delayed]
robotswantdata · 2h ago
Codex CLI IDE just works, very impressed with the quality. If you tried it a while back and didn’t like it, try it again via the vscode extension generous usage included with plus.
Ditched my Claude code max sub for the ChatGPT pro $200 plan. So much faster, and not hit any limits yet.
steinvakt2 · 1h ago
Im using Cursor with the $20 plan and hit rate limits after 15 days (so im paying extra the rest of the month). What do you recommend I do?
robotswantdata · 1h ago
You could get two plus accounts? Or maybe a business account with two seats?
The $200 pro feels good value personally.
poszlem · 1h ago
Wait, what? They now allow claude code like subscription instead of the API too?
robotswantdata · 1h ago
Yes for at least a month. Download the vscode extension and sign in with ChatGPT
Tiberium · 1h ago
Yes, just do "codex login" and it'll use your ChatGPT subscription.
stopachka · 39m ago
Very impressive. I've been working on a shared background presence animation, and have been testing out Claude and Codex. (By shared presence, I mean imagine a page's background changing based on where everyone's cursor is)
Both were struggling yesterday, with Claude being a bit ahead. Their biggest problems came with being "creative" (their solutions were pretty "stock"), and they had trouble making the simulation.
Tried the same problem on Codex today. The design it came up with still felt a bit lackluster, but it did _a lot_ better on the simulation.
simianwords · 30m ago
The code review thing might be my favorite UX for AI based development. Largely stays out of your way and provides good comments.
I’m imagining if it can navigate the codebase and modify tests - like add new cases or break the tests by changing a few lines. This can actually verify if the tests were doing actual assertions and being useful.
Thorough reviewing like this probably benefits me the most - more than AI assisted development.
simianwords · 35m ago
OpenAI is starting its new era of specialized models. Guess they gave up on a monolithic model approach
Tiberium · 4h ago
Only an 1.7% upgrade on SWE-Bench compared to GPT-5, but 33.9 vs 51.3% on their internal code refactoring benchmark. This seems like an Opus 4.1-like upgrade, which is nice to see and means they're serious about Codex.
ianbutler · 2h ago
I just want the codex models in the API, I won’t touch them until then.
And before someone says it, I do happen to have my own codex like environment complete with development containers, browser, github integration, etc.
And I'm happy to pay a mint for access to the best models.
greyb · 1h ago
They've said it's coming:
>For developers using Codex CLI via API key, we plan to make GPT‑5-Codex available in the API soon.
ianbutler · 1h ago
I saw that, but soon doesn’t inspire confidence, and is easy to overlook if they don’t. They didn’t with the previous Codex model.
alvis · 4h ago
It's interest to see this quote:
`for the bottom 10% of user turns sorted by model-generated tokens (including hidden reasoning and final output), GPT‑5-Codex uses 93.7% fewer tokens than GPT‑5`
It sounds like it can make simple tasks much more correct. It's impressive to me. Today coding agent tends to pretend they're working hard by generating lots of unnecessary code. Hope it's true
bn-l · 2h ago
This is my issue with gpt-5. If you use the low or medium reasoning it’s garbage. If you use high, it’ll think for up to five minutes on something dead simple.
srcreigh · 25m ago
Can you be more specific about what type of code you're talking about, and what makes it garbage?
I'm happy with medium reasoning. My projects have been in Go, Typescript, React Dockerfiles stuff like that. The code almost always works, it's usually not "Clean code" though.
Topfi · 2h ago
One major improvement I have seen today, even before I saw the announcement, was that the model is far more reliable in using the Task Completion interface to communicate what stage of the prompt is being implemented. Previously this was only shown sparingly (especially in the first few weeks) and if, it didn't properly tick tasks, simply jumping from the first to completion at the end. Now this works very reliably and I do like this improvement, but if I didn't know better, would have suspected this was merely the result of a system prompt change, considering GPT-5 adherence being very solid in my experience, this should have been fixable without a tuned model. Nevertheless, I like this improvement (arguably fix of a previously broken feature).
Beyond that, purely anecdotal and subjective, but this model does seem to do extensive refactors with semi precise step-by-step guidance a bit faster (comparing GPT-5 Thinking (Medium) and GPT-5 Codex (Medium)), though adherence to prompts seems roughly equivalent between the two as of now. In any case, I really feel they should consider a more nuanced naming convention.
New Claude Sonnet 3.7 was a bit of a blunder, but overall, Anthropic has their marketing in tight order compared to OpenAI. Claude Code, Sonnet, Opus, those are great, clear differentiating names.
Codex meanwhile can mean anything from a service for code reviews with Github integration to a series of dedicated models going back to 2021.
Also, while I do enjoy the ChatGPT app integration for quick on-the-go work made easier with a Clicks keyboard, I am getting more annoyed by the drift between Codex VSCode, Codex Website and Codex in the ChatGPT mobile app. The Website has a very helpful Ask button, which can also be used to launch subtasks via prompts written by the model, but such a button is not present in the VSCode plugin, despite subtasks being something you can launch from the VSCode plugin if you have used Ask via the website first. Meanwhile, the iOS app has no Ask button and no sub task support and neither the app, nor VSCode plugin show remote work done beyond abbreviations, whereas the web page does show everything. Then there are the differences between local and remote via VSCode and the CLI, ... To people not using Codex, this must sound insane and barely understandable, but it seems that is the outcome of spreading yourself across so many fields. CLI, dedicated models, VSCode plugin, mobile app, code review, web page, some like Anthropic only work on one or two, others like Augment three, but no one else does that much, for better and worse.
I like using Codex, but it is a mess with such massive potential that needs a dedicated team lead whose only focus is to untangle this mess, before adding more features. Alternatively, maybe interview a few power user on their actual day to day experience, those that aren't just in one, but are using multiple or all parts of Codex. There is a lot of insight to be gained from someone who has an overview off the entire product stack, I think. Sending out a questionnaire to top users would be a good start, I'd definitely answer.
j45 · 1h ago
Has anyone hit any programming usage limits with the ChatGPT 5 Pro account?
robotswantdata · 46m ago
None yet, feels unlimited.
Huge repos too
incomingpain · 4h ago
Still waiting on codex cli to support lm studio.
NitpickLawyer · 1h ago
? Isn't lmstudio API openai compatible? Codex cli already supports 3rd party models, you have to edit the config yaml file, and you can add many model providers.
trilogic · 55m ago
Is so good that need advertising, LOL. (Anthropic enters chat)...
Jokes apart, I guess it is good (cause I cant afford it to try). (Meta enters chat)...
Deepseek R2 enters chat and drop mike. Anthtopic, Meta, OpenAi, Google, leaves chat and call Trump for another Dinner.
No really Jokes apart, ChatGPT and Trump are awesome, all works and keeps getting better by the day.
laidoffamazon · 41m ago
I agree because I don’t want to get fired or get audited
SWE-bench performance is similar to normal gpt-5, so it seems the main delta with `gpt-5-codex` is on code refactors (via internal refactor benchmark 33.9% -> 51.3%).
As someone who recently used Codex CLI (`gpt-5-high`) to do a relatively large refactor (multiple internal libs to dedicated packages), I kept running into bugs introduced when the model would delete a file and then rewrite in (missing crucial or important details). My approach would have been to just the copy the file over and then make package-specific changes, so maybe better tool calling is at play here.
Additionally, they claim the new model is more steerable (both with AGENTS.md and generally). In my experience, Codex CLI w/gpt-5 is already a lot more steerable than Claude Code, but any improvements are welcome!
[0]https://github.com/openai/codex/blob/main/codex-rs/core/gpt_...
[1]https://github.com/openai/codex/blob/main/codex-rs/core/prom...
Prompting is different, but in a good way.
With Claude Code, you can use less prompting, and Claude will get token happy and expand on your request. Great for greenfield/vibing, bad for iterating on existing projects.
With Codex CLI, GPT-5 seems to handle instructions much more precisely. It won't just go off on it's own and do a bunch of work, it will do what you ask.
I've found that being more specific up-front gets better results with GPT-5, whereas with Claude, being more specific doesn't necessarily stop the eagerness of it's output.
As with all LLMs, you can't compare apples to oranges, so to clarify, my experiences are primarily with Typescript and Rust codebases.
It seems about half my sessions quickly become "why did you do that? rip __ out and just do ___". Then again, most of the other sessions involve Codex correctly inferring what I wanted without having to be so specific.
SWE-bench is a great eval, but it's very narrow. Two models can have the same SWE-bench scores but very different user experiences.
Here's a nice thread on X about the things that SWE-bench doesn't measure:
https://x.com/brhydon/status/1953648884309536958
https://nitter.net/brhydon/status/1953648884309536958
GPT-5 may underwhelm with the same sparse prompt, as it seems to do exactly what's asked, not more
You can still "fully vibe" with GPT-5, but the pattern works better in two steps:
1. Plan (iterate on high-level spec/PRD, split into actions)
2. Build (work through plans)
Splitting the context here is important, as any LLM will perform worse as the context gets more polluted.
For people that have not tried it in say ~1 month, give Codex CLI a try.
Ditched my Claude code max sub for the ChatGPT pro $200 plan. So much faster, and not hit any limits yet.
The $200 pro feels good value personally.
Both were struggling yesterday, with Claude being a bit ahead. Their biggest problems came with being "creative" (their solutions were pretty "stock"), and they had trouble making the simulation.
Tried the same problem on Codex today. The design it came up with still felt a bit lackluster, but it did _a lot_ better on the simulation.
I’m imagining if it can navigate the codebase and modify tests - like add new cases or break the tests by changing a few lines. This can actually verify if the tests were doing actual assertions and being useful.
Thorough reviewing like this probably benefits me the most - more than AI assisted development.
And before someone says it, I do happen to have my own codex like environment complete with development containers, browser, github integration, etc.
And I'm happy to pay a mint for access to the best models.
>For developers using Codex CLI via API key, we plan to make GPT‑5-Codex available in the API soon.
It sounds like it can make simple tasks much more correct. It's impressive to me. Today coding agent tends to pretend they're working hard by generating lots of unnecessary code. Hope it's true
I'm happy with medium reasoning. My projects have been in Go, Typescript, React Dockerfiles stuff like that. The code almost always works, it's usually not "Clean code" though.
Beyond that, purely anecdotal and subjective, but this model does seem to do extensive refactors with semi precise step-by-step guidance a bit faster (comparing GPT-5 Thinking (Medium) and GPT-5 Codex (Medium)), though adherence to prompts seems roughly equivalent between the two as of now. In any case, I really feel they should consider a more nuanced naming convention.
New Claude Sonnet 3.7 was a bit of a blunder, but overall, Anthropic has their marketing in tight order compared to OpenAI. Claude Code, Sonnet, Opus, those are great, clear differentiating names.
Codex meanwhile can mean anything from a service for code reviews with Github integration to a series of dedicated models going back to 2021.
Also, while I do enjoy the ChatGPT app integration for quick on-the-go work made easier with a Clicks keyboard, I am getting more annoyed by the drift between Codex VSCode, Codex Website and Codex in the ChatGPT mobile app. The Website has a very helpful Ask button, which can also be used to launch subtasks via prompts written by the model, but such a button is not present in the VSCode plugin, despite subtasks being something you can launch from the VSCode plugin if you have used Ask via the website first. Meanwhile, the iOS app has no Ask button and no sub task support and neither the app, nor VSCode plugin show remote work done beyond abbreviations, whereas the web page does show everything. Then there are the differences between local and remote via VSCode and the CLI, ... To people not using Codex, this must sound insane and barely understandable, but it seems that is the outcome of spreading yourself across so many fields. CLI, dedicated models, VSCode plugin, mobile app, code review, web page, some like Anthropic only work on one or two, others like Augment three, but no one else does that much, for better and worse.
I like using Codex, but it is a mess with such massive potential that needs a dedicated team lead whose only focus is to untangle this mess, before adding more features. Alternatively, maybe interview a few power user on their actual day to day experience, those that aren't just in one, but are using multiple or all parts of Codex. There is a lot of insight to be gained from someone who has an overview off the entire product stack, I think. Sending out a questionnaire to top users would be a good start, I'd definitely answer.
No really Jokes apart, ChatGPT and Trump are awesome, all works and keeps getting better by the day.
It is inevitable.