ANy one have use kimi k2 in cursor

[Hassan22]

Describe the Bug

so everyone knows chinese qwen and moonshot ai models are performing well if not like anthropic claudes, so why cursor dont want its user to use cheap models ,
they have added kimi 2 but its so frustrating slow and stuck in between, so are we forced to use high paid models so we end up our quiries fast and use pay as you go service, why are cursor not taking benefits from chinese models

Steps to Reproduce

its definitely a bug or on purpose that cusror is not paying attention, are we suppose to use claude sonnet 4 everything ? plz use kini 2 hosted in groq we will be thanking you guys so much

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

macos sceqoia 15.5 (24F74)

Does this stop you from using Cursor

No - Cursor works, but with this issue

[yakovw]

Interesting — I didn’t really notice any slowness.
In my case, I had a different strange bug:
it started inserting a line break between every token, which made the output completely unreadable.
But the response speed itself was reasonable.

[Somhairle_Marisol]

+1, and tool invocations often fail to work.

[yakovw]

For me, it fell before he tried to call the instrument… so I don’t know, but there were times when he did succeed.

[deanrie]

Hey, thanks for the report, we’re currently working on optimizing this model in Cursor. Regarding performance, our provider fireworks.ai is expected to increase the model capacity soon.

[Daed]

when will cursor fix the display issue of kimi k2 i’ve waiting for more than a week since 1.2 until 1.3

[yakovw]

This is not a display problem.
The model adds line breaks between tokens
This is something broken in the model