Add support for o3-mini-high

[golosovsky]

Would highly appreciate if o3-mini-high would be added to Cursor.

[deanrie]

Hey, this model is already used in Cursor.

[omarqurioos]

Hey @deanrie ,
when is it expected to have o3-mini work as smoothly as claude with Composer Agent ?

because it doesn’t apply code to files currently

[sk12345]

When I try to use it, I get “The model o3-mini-high does not work with your current plan…”

When I turn on my openai API Key:
{“error”:{“message”:“The model o3-mini-high does not exist or you do not have access to it.”,“type”:“invalid_request_error”,“param”:null,“code”:“model_not_found”}}

Version: 0.45.9

VSCode Version: 1.96.2

Commit: cce0110ca40ceb61e76ecea08d24210123895320

[Apro123]

I am on the pro plan as well. I can use o3-mini. But I have no idea what version of o3-mini it is? I tried manually setting o3-mini-<level> for each high, medium, and low, and they all give me an error message.

[bannawee]

Qwen pls

[sqlguy42]

it’s listed as “o3-mini” in cursor. are you saying that’s the o3-mini-high model?

[seth]

It appears o3-mini works in cursor, but o3-mini-high which is fine tuned on coding is not available yet. it is also not available in open router. any idea when this will be fixed @deanrie ?

Thank you!!!

[dmwyatt]

I don’t think so? o3-mini != o3-mini-high

[webcpu]

o3-mini-high isn’t a model name. Instead, o3-mini-high represents the o3-mini model with a reasoning_effort parameter. For example:
{
model: “o3-mini”,
reasoning_effort: “high”
}
Please support the reasoning_effort parameter. Thanks!

[m2ahmad2020]

Hi, so if I want use o3 mini high is enough to select o3 mini in the Cursor settings? thank you

[lukaTTT]

@sk12345

When I try to use it, I get “The model o3-mini-high does not work with your current plan…”

When I turn on my openai API Key:
{“error”:{“message”:“The model o3-mini-high does not exist or you do not have access to it.”,“type”:“invalid_request_error”,“param”:null,“code”:“model_not_found”}}

I get this same thing. Anyone got to a solution? O3 in cursor isn’t the best i feel like

[danperks]

Hi all, o3-mini is set to high reasoning behind the scenes, but you cannot manually specific the reasoning with model names like o3-mini-high right now!

[Orpheus]

I recommend to indicate that it is the high reasoning mode in

• Cursor – Models
• Update: o3-mini pricing
• and in the Cursor app

Because the ‘o3-mini’ moniker alone is underspecified.

[ysetbon]

Thanks danperks for the update, how can I make sure that o3-mini is truly set as to high by default?

[danperks]

Hey, o3-mini is set to high behind the scenes by us on our back end and is not configurable by a user yet.

[t851029]

+100, big difference between o3-mini-med and o3-mini-high. It’s fine to have just o3-mini in the IDE, but as Orpheus mentioned, having the details on the models page would be appreciated. Copy change = 1 story point

[lukaTTT]

Guys i still have a feeling that o3-mini for me is mid or even low o3-mini. o1 is giving me significantly better results and that shouldn’t be the case if it’s o3-mini high. I have o3-mini-high in my menu but can’t use it.

[abdo]

I agree, the output from ‘o3-mini’ model on cursor is much lesser intelligent than response from ‘o3-mini-high’ on ChatGPT or API calls. (e.g. For migrating @nuxt/content from v2 to v3, the same model on ChatGPT/API calls is giving different output from the calls within Cursor).