Feature Request: Support Custom AI API Endpoints

[Hydro]

Hi PDF-XChange Team,

The new AI Assistant in PDF-XChange Editor V11 is a very useful feature. It can summarize PDFs, answer questions, extract key information, and help review documents.

However, I hope the AI Assistant can support **custom AI API endpoints**, especially **OpenAI-compatible APIs**.

For example, users could manually configure:

* API base URL
* API key
* Model name
* OpenAI-compatible API format

This would allow users to connect local models or third-party models such as DeepSeek, Qwen, Ollama, LM Studio, vLLM, and private enterprise AI services.

This feature would be very helpful for users who need better privacy, lower cost, local deployment, or specialized model support.

I hope PDF-XChange Editor can add a “Custom API Provider” option in the AI Assistant settings.

Thank you for considering this request.

[Daniel - PDF-XChange]

Hello, Hydro

It is on the list to be added. Slightly more complex than it seems at face value, and for the moment our goal is to thoroughly test the models we do support, before opening the floodgates to numerous other items.

Kind regards,

[Hydro]

Hi Daniel,

Thank you for the reply. It is good to know that custom API endpoints are already on the list.

I would like to add one related issue about AI Assistant network configuration.

When using PDF-XChange Editor V11 in an internal network environment, the AI Assistant fails with:

“The request could not be completed.”

The same AI provider and API key work normally on the same computer at home, and they also work in another application in the internal network.

After testing, I found that enabling the system proxy does not allow PDF-XChange Editor AI Assistant to connect. However, when using a network-level tunneling mode, the AI Assistant works correctly.

This makes me wonder whether the AI Assistant reads the operating system’s proxy settings correctly, or whether its AI API requests bypass the system proxy.

Could you please clarify whether AI Assistant currently supports the system proxy? Also, would it be possible to add explicit proxy settings for AI services in the future?

This would be useful for users in corporate, university, or other internal network environments where external AI API access requires a proxy.

Thank you.

[Daniel - PDF-XChange]

Hello, Hydro

Thank you for the report, I have not observed this myself, but I am raising this to the Dev team fore research and resolution!

Kind regards,