感谢您辛勤的付出！Thank you for this great module. ：）
The following content was edited and summarized using AI.
Is your feature request related to a problem? Please describe.
Currently, this module is specifically designed to work with a local Ollama instance. While Ollama is great, running Large Language Models (LLMs) locally consumes significant CPU/GPU and RAM resources. For many server owners running AzerothCore on limited hardware, this can lead to game lag or even server crashes when the AI is generating responses.
Describe the solution you'd like
I would like to suggest adding support for third-party API endpoints or aggregators (such as NewAPI, OneAPI, or direct OpenAI-compatible services).
This would require a few additions to the configuration and the HTTP request logic:
- Base URL Configuration: Allow users to define a custom endpoint (e.g.,
https://api.your-provider.com/v1/chat/completions).
- API Key Support: Add a configuration field for an
Authorization header (Bearer YOUR_API_KEY).
- OpenAI Protocol Compatibility: Since NewAPI and most providers use the OpenAI standard format, the module might need an option to switch between Ollama’s JSON structure and the OpenAI
/v1/chat/completions structure.
Describe alternatives you've considered
One could manually edit the scripts to change the URL and headers, but having this integrated into the mod_ollama_chat.conf would make it much more accessible for the community.
Additional context
By supporting external APIs:
- Server Performance: The heavy lifting of AI inference is offloaded to a remote service, keeping the game server's CPU/RAM usage near zero.
- Flexibility: Users can choose cheaper or more powerful models (like DeepSeek, GPT-4o-mini, etc.) that are often more cost-effective and smarter than small local models.
- Stability: It prevents the game thread from being impacted by local hardware spikes during AI generation.
Thank you for this great module, and I look forward to hearing your thoughts!
感谢您辛勤的付出！Thank you for this great module. ：）
The following content was edited and summarized using AI.
Is your feature request related to a problem? Please describe.
Currently, this module is specifically designed to work with a local Ollama instance. While Ollama is great, running Large Language Models (LLMs) locally consumes significant CPU/GPU and RAM resources. For many server owners running AzerothCore on limited hardware, this can lead to game lag or even server crashes when the AI is generating responses.
Describe the solution you'd like
I would like to suggest adding support for third-party API endpoints or aggregators (such as NewAPI, OneAPI, or direct OpenAI-compatible services).
This would require a few additions to the configuration and the HTTP request logic:
https://api.your-provider.com/v1/chat/completions).Authorizationheader (Bearer YOUR_API_KEY)./v1/chat/completionsstructure.Describe alternatives you've considered
One could manually edit the scripts to change the URL and headers, but having this integrated into the
mod_ollama_chat.confwould make it much more accessible for the community.Additional context
By supporting external APIs:
Thank you for this great module, and I look forward to hearing your thoughts!