1
[Feature Request] Expose LLM extra parameters in the AI settings UI
Source: paperless-ngx/paperless-ngx#14648 · opened by @Tranquillius
Description Please add PAPERLESS_AI_LLM_EXTRA_PARAMS as a field in the AI settings UI, next to the backend, endpoint and model fields. Some models reject the default temperature. With claude-sonnet-5 and claude-haiku-5-5 on the OpenAI-compatible Anthropic endpoint, AI suggestions fail with a 502 and the log shows: Error code: 400 - {'error': {'code': 'invalid_request_error', 'message': 'temperature is deprecated for this model.', 'type': 'invalid_request_error', 'param': None}} Setting PAPERLESS_AI_LLM_EXTRA_PARAMS='{"temperature": 1}' fixes this. Because it is environment-only, users who configure everything in the UI have to edit their compose file and restart the container. The failing request also gives no hint that this setting exists. This does not ask Paperless-ngx to support specific models or keep a per-model list. Users could handle such cases themselves from the …
No pledges yet. Be the first to back this.
Comments
Similar requests
[Feature Request] Different keys for LLM vs. LLM embeddings
6 votes · 0 comments
[Feature Request] make "test connection" button for LLM endpoints /add 2nd backend provider for LLMs
1 vote · 0 comments
[Feature Request] AI Suggestions Cache
12 votes · 0 comments
Add support for Ollama's think: false parameter for AI suggestions
1 vote · 0 comments
[Feature Request] Allow overriding the embeddings encoding_format for OpenAI-compatible LLM providers
4 votes · 0 comments
No comments yet.