opencode icon indicating copy to clipboard operation
opencode copied to clipboard

Model Not Exist: When using deepseek-v3.2

Open patrickelectric opened this issue 1 month ago • 7 comments

Description

When using Deepseek-v3.2 it says that model does not exist

OpenCode version

1.0.137

Steps to reproduce

  1. Change model to deepseek-v3.2 and type hi

Screenshot and/or share link

Image

Operating System

Arch (BTW)

Terminal

konsole/tmux

patrickelectric avatar Dec 09 '25 14:12 patrickelectric

This issue might be a duplicate of existing issues. Please check:

  • #2873: Error with model DeepSeek V3 Base (free) - Similar ProviderModelNotFoundError when using deepseek models
  • #3993: CLI run --model fails for namespaced models with : or multi-segment paths - Could affect models with special characters in naming

Feel free to ignore if none of these address your specific case.

github-actions[bot] avatar Dec 09 '25 14:12 github-actions[bot]

Provider?

JunarisAlf avatar Dec 09 '25 15:12 JunarisAlf

Provider?

I just used /models -> deepseek v3.2 -> put my API key

patrickelectric avatar Dec 09 '25 16:12 patrickelectric

Provider?

I just used /models -> deepseek v3.2 -> put my API key

What? It's should be opencode auth login -> choose provider -> enter api key -> select models /models If you not touching opencode.json then probably the database models.dev out of date. Try check actual model id on the provider website then configure manually

JunarisAlf avatar Dec 09 '25 16:12 JunarisAlf

If you can tell me what provider you are using I can help

rekram1-node avatar Dec 09 '25 16:12 rekram1-node

Provider?

I just used /models -> deepseek v3.2 -> put my API key

What? It's should be opencode auth login -> choose provider -> enter api key -> select models /models

opencode auth login

┌  Add credential
│
◇  Select provider
│  DeepSeek
│
◇  Enter your API key
│  ▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪
│
└  Done
Image

If you not touching opencode.json then probably the database models.dev out of date. Try check actual model id on the provider website then configure manually

haa... any idea of how to do that ?

I did run...

 opencode models deepseek --refresh       
Models cache refreshed
deepseek/deepseek-chat
deepseek/deepseek-reasoner
deepseek/deepseek-v3.2

But the problem continues

patrickelectric avatar Dec 09 '25 18:12 patrickelectric

maybe the model id is invalid lemme look in a bit

rekram1-node avatar Dec 09 '25 19:12 rekram1-node

this happened to me with deepseek v3.2 just now. go to /connect select Deepseek, enter api key, select model, say hi. "model not found."

gminorcoles avatar Dec 10 '25 13:12 gminorcoles

based on this the reasoner and chat are already deepseek v3.2:

https://api-docs.deepseek.com/quick_start/pricing

i guess the base option is invalid enitrely so we should remove it

rekram1-node avatar Dec 10 '25 14:12 rekram1-node

Yeah the model was mistakenly added, they only have reasoner and chat, both use deepseek v3.2

rekram1-node avatar Dec 10 '25 17:12 rekram1-node

I removed it from the options, it will take a minute for your caches to sync

rekram1-node avatar Dec 10 '25 17:12 rekram1-node