[v0.14.7] Read tool incorrectly detects model as non-image-capable #3084

Activity

1. rekram1-node commented on Oct 10, 2025

   

   rekram1-node

   on Oct 10, 2025

   Last edited by rekram1-node

   Contributor

   More actions

   I think we need to update models.dev to track it correctly..

   is this a custom claude sonnet 4? I see "Nexus Claude Sonnet 4" in your screenshot

   But the other part looks like normal?

   What provider is this?
2. rekram1-node

   self-assigned this

   on Oct 10, 2025
3. Raviguntakala commented on Oct 10, 2025

   

   Raviguntakala

   on Oct 10, 2025

   ContributorAuthor

   More actions

   I think we need to update models.dev to track it correctly..

   is this a custom claude sonnet 4? I see "Nexus Claude Sonnet 4" in your screenshot

   But the other part looks like normal?

   What provider is this?

   Nexus is my client’s internal API gateway
4. rekram1-node commented on Oct 10, 2025

   

   rekram1-node

   on Oct 10, 2025

   Last edited by rekram1-node

   Contributor

   More actions

   @Raviguntakala can u show your opencode.json if this is a custom model you need to specify the modalities currently...

   ill show you, here is an example:

   {
     "$schema": "https://opencode.ai/config.json",
     "provider": {
       "myprovider": {
         "npm": "@ai-sdk/openai-compatible",
         "name": "My AI ProviderDisplay Name",
         "options": {
           "baseURL": "https://api.myprovider.com/v1"
         },
         "models": {
           "my-model-name": {
             "name": "My Model Display Name",
             "modalities": {
               "input": ["text", "image"],
               "output": ["text"]
             }
           }
         }
       }
     }
   }
   

   👍React with 👍2Reacted by Victor Nogueira and fefa4ka

   🎉React with 🎉2Reacted by Victor Nogueira and fefa4ka
5. Raviguntakala commented on Oct 11, 2025

   

   Raviguntakala

   on Oct 11, 2025

   Last edited by Raviguntakala

   ContributorAuthor

   More actions

   "modalities": {
   "input": ["text", "image"],
   "output": ["text"]

   Thanks @rekram1-node! After adding modalities, it is working.

   
6. rekram1-node commented on Oct 11, 2025

   

   rekram1-node

   on Oct 11, 2025

   Contributor

   More actions

   sweet
7. rekram1-node

   closed this as completedon Oct 11, 2025
8. timblaktu

   added a commit that references this issue on Feb 3, 2026

   Fix OpenCode vision support by adding modalities configuration to qwe…

   ...

   ad71afc