Compatibility with gpt-5 codex
Pre-submit Checks
- [x] I have searched Warp feature requests and there are no duplicates
- [x] I have searched Warp docs and my feature is not there
Describe the solution you'd like?
Any chance of getting warp to use got-5 codex?
Is your feature request related to a problem? Please describe.
No response
Additional context
No response
Operating system (OS)
macOS
How important is this feature to you?
4
Warp Internal (ignore) - linear-label:39cc6478-1249-4ee7-950b-c428edfeecd1
None
Any plans of this happening soon?
gpt-5-codex is not yet available via API
For developers using Codex CLI via API key, we plan to make GPT‑5-Codex available in the API soon.
It's on the roadmap and will be tested once we can get API access. (Assuming it passes testing/evaluation, we will add it as an option)
It's out there now https://platform.openai.com/docs/models/gpt-5-codex
It's on the roadmap and will be tested once we can get API access. (Assuming it passes testing/evaluation, we will add it as an option)
Please don't implement it like it is in cursor where it routes and decides which thinking level to use. Allow us to select thinking and reasoning level (high, medium, low) like in codex.
@Capedbitmap unfortunately as I understood it gives medium level only in current endpoint however it seems good anyways. @dannyneira approx how long it takes to get thru the testing/eval? :)
Just curious, why was Sonnet 4.5 added almost immediately to Warp and we still do not have Codex? Despite Anthropic's claims, Codex remains the far superior coding model in my opinion, and the lack of availability in Warp precludes me from using Warp full-time and subscribing to Turbo / Lightspeed. Not being sarcastic, genuine question. If the answer is that demand for Sonnet 4.5 is / was expected to be significantly higher than Codex, I totally get the prioritization call.
I'll still use Warp terminal but downgrading from lightspeed to base level due to the quality of results i'm seeing from codex high. I hope the integration goes well tho and can use full warp in the future.
Just curious, why was Sonnet 4.5 added almost immediately to Warp and we still do not have Codex? Despite Anthropic's claims, Codex remains the far superior coding model in my opinion, and the lack of availability in Warp precludes me from using Warp full-time and subscribing to Turbo / Lightspeed. Not being sarcastic, genuine question. If the answer is that demand for Sonnet 4.5 is / was expected to be significantly higher than Codex, I totally get the prioritization call.
Likely due to how they are integrated, the warp agent seems to be built and optimized around sonnet 4, before gtp-5 was launched, so it makes sense that sonnet 4.5 would be an easier integration. Yeah opus seems better than sonnet 4.5 and codex high quite a bit better at least in my results. The code the Claude models generate needs to be audited far more carefully.
When we are going to see codex model?
I heard that Warp are 'still deciding whether GPT-5-Codex is worth adding' on Warp discord channel. To me this sounds disingenuous. Maybe it is inconvenient, maybe it is more work to add than Claude 4.5 Sonnet. But I quote Warps own website here: Our agent always includes access to the top models even as they change, and we will continue to improve the behind-the-scenes context and prompt engineering that makes Warp’s agent the most powerful around.
--> Well gpt-5 changed, and now you are only offering the 2nd tier of it. --> GPT-5-Codex is superior in GPT-5 in all areas and still better than Claude 4.5 Sonnet at code refactoring. In other words while Claude 4.5 Sonnet is superb for daily coding, the best model yet for the bulk of coding, GPT-5-Codex is still the overall leader for the toughest issues. So listen to your users. Add it. We want it. And if we don't get it, some users will definitely leave your platform. I'm pretty sure of that.
ive gone back to just paying the frontier model makers directly so I can get access to gpt5-codex. would love to come back to warp but its not worth it without this model.
When will gpt5-codex be available in Warp? Is there a specific date or plan?
Codex became GA, so even this can't be the reason anymore :)
Hi Folks, we're testing Codex in place of GPT-5, so thanks for your patience while we work on this, folks its for sure on our radar and being worked on.
Hi Folks, we're testing Codex in place of GPT-5, so thanks for your patience while we work on this, folks its for sure on our radar and being worked on.
Why "in place of"? Why would you remove gpt-5 if you added codex? Just add it alongside.
I think alongside and possibly replacing GPT-5, but in general, we are working on Codex being added in some way, shape, or form.
+1 for this one, quick please
+1 cant wait to gets my hands dirty using gpt codex in Warp my fav ai tool ever... @dannyneira
the definition of own-goal but ofc it needs "evaluation" right? :)
Why did it take hours to implement Haiku, but we've been waiting for Codex since last month? I don't feel you guys are being transparent enough of why this is taking so long.
We are actively working on codex support and trying to work out some bug (i.e. wasting credits) with the system prompt.
Thanks for your patience folks!
Is there any realistic ETA when we will be able to use Codex in Warp? @dannyneira There are many of us following this closely—would appreciate any update on a concrete timeline!
Sorry but there is no firm eta a this time. We will post when there are updates to share. Thanks for your understanding!
@dannyneira any updates?
I saw this coming. This is why I cancelled my subscription. Reading between the lines I knew GPT-5-codex would not drop any time soon on Warp. The non-committal responses were deafening from the start, and are still getting louder. I hope you guys get it before Christmas!
I mean 70% of users will unsubscribe and leave to CC/Codex/Gemini because these wrappers became a joke with their pricing. Who the f@ck wants to pay 400$ for what you get in CC or Codex and with the latest models and non constrained context like here... Good luck for them once the yearly plan finishes I just wave goodbye to this solution for sure.
@eduardosantos1989 - It's with no great satisfaction for me to say this but I think it's just a matter of time before the 'big boys' who create the best models e.g. Anthropic, OpenAI, Google, xAI also provide the best tools for the job and take away the business model these IDE middlemen are providing. Because the added value is getting thinner over time. There's only so much ingenuity you can have in a certain budget, and they simply have more resources to throw at the problem, not to mention the best talent.
I hope Codex versions are arriving soon. I find myself using codex CLI more than Warp lately due to this. In addition, the new Build pricing plan is my main reason to consider cancelling my subscription. Sorry, guys.
There are some challenges of making GPT-5-codex work with Warp's harness.
To anyone else interested in this feature, please add a :+1: to the original post at the top to signal that you want this feature, and subscribe if you’d like to be notified.