[GH-ISSUE #2060] Ollama: embedding model does not support chat #1282

Closed
opened 2026-03-02 11:56:15 +03:00 by kerem · 1 comment
Owner

Originally created by @qdii on GitHub (Oct 22, 2025).
Original GitHub issue: https://github.com/karakeep-app/karakeep/issues/2060

Describe the Bug

On version v.0.27.1, inference job fails with the following error:

2025-10-22T07:04:17.331Z error: [inference][2399] inference job failed: ResponseError: "nomic-embed-text" does not support chat                                                                              
ResponseError: "nomic-embed-text" does not support chat                                                                                                                                                      
    at checkOk (file:///app/apps/workers/dist/index.js:59583:8)                                                                                                                                              
    at process.processTicksAndRejections (node:internal/process/task_queues:105:5)                                                                                                                           
    at async post (file:///app/apps/workers/dist/index.js:59634:2)                                                                                                                                           
    at async Ollama.processStreamableRequest (file:///app/apps/workers/dist/index.js:59733:22)                                                                                                               
    at async OllamaInferenceClient.runModel (file:///app/apps/workers/dist/index.js:69526:26)                                                                                                                
    at async OllamaInferenceClient.inferFromText (file:///app/apps/workers/dist/index.js:69570:10)                                                                                                           
    at async inferTagsFromText (file:///app/apps/workers/dist/index.js:86705:9)                                                                                                                              
    at async inferTags (file:///app/apps/workers/dist/index.js:86712:49)                                                                                                                                     
    at async runTagging (file:///app/apps/workers/dist/index.js:86795:15)                                                                                                                                    
    at async Object.runOpenAI [as run] (file:///app/apps/workers/dist/index.js:86860:4)  

I'm using a self-hosted ollama instance with nomic-embed-text as a model.

I presume this happens because the /chat endpoint is being called instead of /embed (see https://github.com/ollama/ollama/blob/main/docs/api.md#generate-embeddings).

Steps to Reproduce

  1. Go to Admin settings
  2. Click on "Regenerate AI tags for Failed Bookmarks Only"

Expected Behaviour

Inference job succeeds

Screenshots or Additional Context

No response

Device Details

Kubernetes cluster

Exact Karakeep Version

v0.27.1

Have you checked the troubleshooting guide?

  • I have checked the troubleshooting guide and I haven't found a solution to my problem
Originally created by @qdii on GitHub (Oct 22, 2025). Original GitHub issue: https://github.com/karakeep-app/karakeep/issues/2060 ### Describe the Bug On version v.0.27.1, inference job fails with the following error: ``` 2025-10-22T07:04:17.331Z error: [inference][2399] inference job failed: ResponseError: "nomic-embed-text" does not support chat ResponseError: "nomic-embed-text" does not support chat at checkOk (file:///app/apps/workers/dist/index.js:59583:8) at process.processTicksAndRejections (node:internal/process/task_queues:105:5) at async post (file:///app/apps/workers/dist/index.js:59634:2) at async Ollama.processStreamableRequest (file:///app/apps/workers/dist/index.js:59733:22) at async OllamaInferenceClient.runModel (file:///app/apps/workers/dist/index.js:69526:26) at async OllamaInferenceClient.inferFromText (file:///app/apps/workers/dist/index.js:69570:10) at async inferTagsFromText (file:///app/apps/workers/dist/index.js:86705:9) at async inferTags (file:///app/apps/workers/dist/index.js:86712:49) at async runTagging (file:///app/apps/workers/dist/index.js:86795:15) at async Object.runOpenAI [as run] (file:///app/apps/workers/dist/index.js:86860:4) ``` I'm using a self-hosted ollama instance with `nomic-embed-text` as a model. I presume this happens because the `/chat` endpoint is being called instead of `/embed` (see https://github.com/ollama/ollama/blob/main/docs/api.md#generate-embeddings). ### Steps to Reproduce 1. Go to Admin settings 2. Click on "Regenerate AI tags for Failed Bookmarks Only" ### Expected Behaviour Inference job succeeds ### Screenshots or Additional Context _No response_ ### Device Details Kubernetes cluster ### Exact Karakeep Version v0.27.1 ### Have you checked the troubleshooting guide? - [x] I have checked the troubleshooting guide and I haven't found a solution to my problem
kerem 2026-03-02 11:56:15 +03:00
Author
Owner

@MohamedBassem commented on GitHub (Oct 26, 2025):

from the error message, this seems to imply that you're using that model as your INFERENCE_TEXT_MODEL, and as the error message says, this is an embedding model and not a chat model. INFERENCE_TEXT_MODEL needs to be a chat model.

<!-- gh-comment-id:3448627298 --> @MohamedBassem commented on GitHub (Oct 26, 2025): from the error message, this seems to imply that you're using that model as your `INFERENCE_TEXT_MODEL`, and as the error message says, this is an embedding model and not a chat model. `INFERENCE_TEXT_MODEL` needs to be a chat model.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
starred/karakeep#1282
No description provided.