Error: "this model does not support image input" with paid providers (GitHub Copilot, Anthropic, etc.) but works with free models #16468

Activity

1. NeuralLabs83

   added

   bugSomething isn't workingSomething isn't working

   on Mar 7, 2026
2. github-actions

   assigned

   rekram1-node

   on Mar 7, 2026
3. github-actions

   added

   coreAnything pertaining to core functionality of the application (opencode server stuff)Anything pertaining to core functionality of the application (opencode server stuff)

   windows

   on Mar 7, 2026
4. github-actions commented on Mar 7, 2026

   

   github-actionsBot

   on Mar 7, 2026 – with GitHub Actions

   Contributor

   More actions

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

   • [Bug] look_at tool cannot read image files - "this model does not support image input" even with vision-capable model #16184: Same error message "this model does not support image input" with vision-capable models
   • model "claude-opus-4.6" not supported for vision #14289: Claude Opus 4.6 not supported for vision — same provider/model combination
   • Read tool cannot pass image data to vision-capable models #15728: Read tool cannot pass image data to vision-capable models
   • [ModelScope] Custom Provider with Qwen Model Cannot Recognize Images - "Image Input Not Supported" #15040: Custom provider with vision-capable model getting "Image Input Not Supported"
   • Unable to read images for some models #5359: Unable to read images for some models

   If your issue is distinct (e.g., specifically about the paid vs free provider distinction), please add more detail to differentiate it from the above.
5. mtomas7 commented on Mar 10, 2026

   

   mtomas7

   on Mar 10, 2026

   More actions

   Please see if this config tip will help: #15728 (comment)
6. thdxr

   removed

   bugSomething isn't workingSomething isn't working

   windows

   coreAnything pertaining to core functionality of the application (opencode server stuff)Anything pertaining to core functionality of the application (opencode server stuff)

   on May 3, 2026
7. botlong commented on May 18, 2026

   

   botlong

   on May 18, 2026

   More actions

   Still hitting this on v1.15.4 (Windows). Custom provider using @ai-sdk/openai with baseURL override — model is gpt-5.5 which is multimodal, but OpenCode reports capabilities.input.image: false.

   Tried every config-level workaround I could find:

   • Adding vision: true to the model definition — ignored
   • Adding capabilities: { input: { image: true } } — ignored
   • Adding attachment: true — sets capabilities.attachment: true but input.image stays false
   • Switching npm adapter from @ai-sdk/openai-compatible to @ai-sdk/openai — no change
   • Using a model ID that OpenCode recognizes (gpt-4o) under the custom provider — still false

   The capability registry appears to be entirely hardcoded in the Go binary and ignores user config overrides. This makes all custom/proxy providers unable to use image input regardless of the model's actual capabilities.

   Would love to see either:

   1. A config-level capabilities override that the registry respects, or
   2. The registry updated to include gpt-5.5 (released by OpenAI, multimodal)

   Thanks for the great project otherwise!

   👀React with 👀1Reacted by r000bin
8. rekram1-node commented on May 18, 2026

   

   rekram1-node

   on May 18, 2026

   Collaborator

   More actions

   There is no "go binary" if u are using a custom model u can define it like so:

   "provider": {
       "my-provider": {
         "models": {
           "my-model": {
             "modalities": {
               "input": ["text", "image", "pdf"],
               "output": ["text"],
             },
           },
         },
       },
     },
   

   👍React with 👍5Reacted by skipp, Valery Lourie, Hello, Naoby Prawira and Chenhao (Eric) Xu
9. rekram1-node

   closed this as not plannedon May 18, 2026
10. valthebald commented on Jun 25, 2026

    

    valthebald

    on Jun 25, 2026

    More actions

    @rekram1-node you saved my day!