🐛 [Bug]: http 401 error when I tried to change the gemini text embedding model. 404 error when submitting
Archon Version
v 0.1.0
Bug Severity
🟡 Medium - Affects functionality
Bug Description
I attempted to change the gemini text embeddings to the current latest model (there is no choice of models) when I tried to input a pdf i received a http 401 error. I tried it again and got the same resulre. I tried to submit the bug through the bug feature within archon and got a 404 error at github.
Steps to Reproduce
- go to settings
- go to RAG setting and select embeddings.
- type any change to the embedding model number.
- save
- ingest pdf
- http 401 error no import
Expected Behavior
it should have successfully ingested the pdf
Actual Behavior
http 401 error
Error Details (if any)
Affected Component
🔍 Knowledge Base / RAG
Browser & OS
Version 141.0.7390.76 (Official Build) (arm64) mac tahoe 26.0.1
Additional Context
No response
Service Status (check all that are working)
- [ ] 🖥️ Frontend UI (http://localhost:3737)
- [ ] ⚙️ Main Server (http://localhost:8181)
- [ ] 🔗 MCP Service (localhost:8051)
- [ ] 🤖 Agents Service (http://localhost:8052)
- [ ] 💾 Supabase Database (connected)
isnt there only
gemini-embedding-001 ?
I know google had text-embedding-004 and below (doesn't make sense to use this though since we have gemini and it stores in a bigger dimension)
gemini-embedding-001 triggers the same 401 error
gemini-embedding-001 triggers the same 401 error
weird works for me just fine
super weird. I just tried it again and got the same errors.
super weird. I just tried it again and got the same errors.
the 401 error is an authentication error, this involves gemini directly, have you tried a new api key?
This is what google showed me regarding the 401
A "401 Unauthorized" error when interacting with the Google Gemini API or platform indicates an authentication failure. This means the request sent to the Gemini service lacks valid credentials or the provided credentials are insufficient for the requested action.
I just switched to openai, no problems there.