Deepseek/deepseek-chat - 400 Model Not Exist error

[Grelo4ka]

Which API Provider are you using?

OpenRouter

Which Model are you using?

deepseek-chat

What happened?

Model deepseek-chat does not seem to work after the latest update.

Steps to reproduce

1. Choose deepseek as a provider
2. Choose deepseek-chat as the model

Relevant API REQUEST output

Additional context

No response

[mrubens]

Weird! What if you switch from chat to R1 and then back again in the settings?

[Grelo4ka]

Weird! What if you switch from chat to R1 and then back again in the settings?

Yep, switching to R1 and then back to chat indeed makes it work

[youpsla]

Got the bug too. Switching from R1 back to chat does not work at first. Works after few try ....

[mrubens]

Hmm I'll take a closer look. Very likely it's related to a change I made earlier today. Sorry for the hassle!

[ditti-o]

Both work for me via the Requesty router...
chat: deepseek/deepseek-chat
r1: cline/deepseek-reasoner:alpha

[Grelo4ka]

Both work for me via the Requesty router... chat: deepseek/deepseek-chat r1: cline/deepseek-reasoner:alpha

You can run it directly via the OpenAI compatible format, so it's not that big of a deal - and also better to use it this way directly. Matt also provided a working and easy solution to bypass the issue. Anyways, no need to use any middlemen like the one you've mentioned - they charge arbitrary fees and/or collect your data. The issue is regarding the Deepseek provider only, not any other providers, as I've specified

[ditti-o]

Thanks @Grelo4ka, that's very useful!

For what it's worth, I was also concerned about those things, so I actually reached out to them:

• You have a config switch that guarantees nothing is stored on their servers
• They do charge a small extra (5%), but I have a single API key and I don't hit rate limits

And btw, deepseek-reasoner didn't work out of the box for me because of the temperature argument. But it did via their API, so I've got a new toy 😆

[Grelo4ka]

Thanks @Grelo4ka, that's very useful!

For what it's worth, I was also concerned about those things, so I actually reached out to them:

* You have a config switch that guarantees nothing is stored on their servers

* They do charge a small extra (5%), but I have a single API key and I don't hit rate limits

And btw, deepseek-reasoner didn't work out of the box for me because of the temperature argument. But it did via their API, so I've got a new toy 😆

Yeah, that's what I figured - you shall use OpenAI compatible option instead or use the fix suggested by Matt and not pay any extra fees at all. Deepseek does not have any rate limits, so your answer is irrelevant.

Given that you've only created your account this month, and have no public repositories nor commits at all, you seem to promote the service and have some interest in that or an affiliation to the website that you've linked.

Also, using another provider is not the solution of a problem. Please, keep the discussion relevant instead of promoting third-party services

[alex595ururu]

Пытаюсь подключиться к DeepSeek

и получаю ошибку

если я обращаюсь по API
то все ок

[simbav911]

Deep Seek is not working when we connect over SSH. However, it works fine on the local machine."

[laurencebush]

R1 is called deepseek-reasoner, it should be viewable now, as I've got it all good.