#The model hit an error

70 messages · Page 1 of 1 (latest)

prime raft
#

Been enjoying making my app with Chef. I've run into the model hit an error repeat message. I searched search.convex.dev/ to no avail. I can't send any message without the response. Has anyone solved this, or am I just stuck?

signal cradle
#

hi. are you using "Auto" mode or did you pick a particular model?

silent sparrow
#

@signal cradle I'm having this issue too. I've tried dozens of times, with Auto, Anthropic, and GPT-4.1. Doesn't seem to matter what prompt I give it (I've tried 3 different ones). I'm stuck and cannot proceed. Could it be that I've exceeded the context window? There's a lot of prior work in there.

signal cradle
#

Maybe. We'll dig into the errors on our side to see what's up and ship more fixes this week.

dire hull
#

Yes, I have the same issue. I tried many times. I changes API keys many times too. But, still the error.

I want to participate in the hackthon, but due to that error, I am unable to develop the web app.

Could the hackathon deadline be extended please?

loud granite
#

I guess it's the related but I face this error the moment I want to use an own API key, in my case for Gemini 2.5. Saving the API key works, so does the API key (tested with another service) but in the chat I receive:

"The model hit an error. Try sending your message again."

Here are some more details @signal cradle:

  • Frontend Error:

    • Error: Failed to process successful response
    • Location: StreamingIndicator.tsx and index.mjs:1127
    • Retried 1+ times with same result.

  • Network Request:

    • URL: GET https://little-cuttlefish-93.convex.cloud/api/check_admin_key
    • Status: 404 Not Found
    • Request Headers: included authorization (token present but not logged)
    • Response Body: empty
    • Referrer: https://dashboard-embedded.convex.dev

  • Timing Observations:

    • Convex container boot took ~355s before reaching READY state.

  • Browser/Environment:

    • Browser: Chrome 135.0.0.0
    • OS: macOS 10.15.7
marsh yarrow
#

I have the same issue, could this be fixed by creating the Credentials in the Gemini API ?
Testing the Gemini API locally seems to work, but the Error you describe suggest that it could be related?

lilac hill
#

I am getting the same issues on Gemini, Antropic and Auto

novel badger
#

I have the same issue using Gemini

gaunt zephyr
#

hi! sorry about this issue. we identified a bug on our side and will have a fix out in the next few hours.

gaunt zephyr
#

thanks for your patience, the fix is out now! let us know if you have any more issues.

marsh yarrow
#

With Gemini, it now often generates && instead of &&

loud granite
#

@gaunt zephyr I can confirm that the reported bug in specific is now resolved, thanks! (only tried for gemini 2.5)

safe rivet
#

I'm on auto mode and still getting the issue

#

first it generates the code and then runs the typescript and when it fails it immediately just shows this error

safe rivet
#

I'm also getting this:

just empty response.

Screenshot_2025-05-10_at_15.59.37.png
wispy flare
#

on auto and on gpt 4.1 and gemeni

safe rivet
#

yeah for me as well, couldn't use throughout the day :((

#

it's literally removing existing code and leaving comment "existing code remains unchanged" 🤣
I don't know what is happening, but I guess you guys probably broke something in Chef 😦

gaunt zephyr
#

hi! can you please explain your issue a bit more? are you running into the error or seeing the empty message?

#

have you tried rewinding and using a slightly different prompt?

safe rivet
#

Hey Jordan!

#

I'm actually getting both, but constantly getting an error message

#

Noticed getting error message only after Typecheck fails, got empty message just randomly few times in a row, but error always after Typecheck fails, and as you could notice Typecheck will always fail since it's strangely updating existing files (assuming it's not changing existing code but in fact completely removing them)

#

and yes, I'm rewinding and tried to use different prompt but it's not helping

gaunt zephyr
#

have you tried reloading the page? we released a fix for the reloading error on friday and you may not be running the latest version of chef.

#

there is also another known issue with overwriting old files incorrectly. i will chat with my coworker on monday.

red forum
#

i am seeing this as well, and i have a hunch. each chat is tied to a single project that chef creates in the convex dashboard. and each chat has a model context limit. if we could have multiple chats per project that might change this? seems similar to what i was dealing with in v0, each chat would create a whole new project and completely replace any code that was written beforehand. i could be wrong on how this is working in chef, ive only been using it for a few days.

safe rivet
gaunt zephyr
swift pagoda
#

hey ive been getting this error since yesterday night UK time. still having same issue after failed typecheck. any ideas folks. tried reloading and all.

Screenshot_2025-05-11_151702.png
gaunt zephyr
swift pagoda
#

Got lots of days of work in there😭

gaunt zephyr
#

okay, this is a bug on our side. we are investigating now. to unblock you for the moment, can you try using gemini 2.5 in the model dropdown? this should get things working for you.

swift pagoda
gaunt zephyr
#

no

swift pagoda
#

Just out of curiosity, what model does it prioritise when its set on auto?

gaunt zephyr
#

also, let me know if the workaround fixes things for you

swift pagoda
#

seems like it's working. Can you pls post here when you guys push a hotfix for auto model. Cheers mate

tender fog
#

I have been encountering this same error for some days now

image.png
#

[ERROR] [Chat] Request failed (retries: {"numFailures":0,"nextRetry":1747054487046}) Error: Overloaded
at onErrorPart (index.mjs:1127:13)
at WG (index.mjs:826:49)
at async $G (index.mjs:932:3)
at async YG (index.mjs:1244:7)
at async index.mjs:306:9 undefined
3StreamingIndicator.tsx:133 Error: Overloaded
at onErrorPart (index.mjs:1127:13)
at WG (index.mjs:826:49)
at async $G (index.mjs:932:3)
at async YG (index.mjs:1244:7)
at async index.mjs:306:9

This is the log in the console when the error happens @gaunt zephyr

red forum
gaunt zephyr
gaunt zephyr
red forum
gaunt zephyr
red forum
#

yes, i ran out of chef tokens a few days ago

tender fog
drowsy brook
#

Does refreshing the page and then sending a message work?

drowsy brook
#

We just released chef with a fix - let us know if you're still seeing this issue! Thanks for your patience and feedback

red forum
#

came back to try chef again, still getting the error. and gemini wont work at all, they are limiting all pro requests. can we use flash somehow? im using api keys

swift pagoda
#

hey I'm getting model error again since this morning. using gemini @gaunt zephyr any ideas mate? //The model hit an error. Try sending your message again.//

swift pagoda
#

prob issue is with Gemini other models are working fine

red forum
#

using my anthropic api key i still see the error

signal cradle
#

is this all on one project, or on any project?

void creek
#

Hi, I get this same erorr even with switching to gemini. Just using your tokens. Chat is not that long. After runing out of free tokens I have purchased the plan to get more and it got stuck.

void creek
swift pagoda
gaunt zephyr
#

hi! sorry you are running into this issue. can you please dm me share link to your project? i can investigate this.

swift pagoda
red forum
#

let us make a new chat without having to make a whole new project please! if i could just start a new chat i think it would solve this issue ❤️

gaunt zephyr
#

we are actively working on this. thanks for your feedback!

balmy gorge
#

I am getting this constantly. I then reload the page, and able to continue. I am using a lot of chef tokens which I am okay with as I am making progress.

gaunt zephyr
#

hi! we are currently investigating this issue. thanks for reporting it.