mendix-ai-assistance-maia

[mdenhartog]

Please use the form below, leaving the prefilled data to help us. Thank you.

Page link: mendix-ai-assistance-maia

Document link: _index.md

My Issue/Suggestion

Running Maia on my local PC with Ollama local installed with 2 GPU's uses always GPU 0 but needs to grap GPU 1 because the second GPU is an NVIDIA GPU. Not able to switch to the correct GPU. Local running Ollama uses the correct GPU but with Mendix not. So I think the developers has programmed hard that you need to use GPU 0. This is not handy to have .... and makes Maia not usable.

[KONRADS098]

Hi @mdenhartog! Thank you for taking the time to report this. We wanted to clarify how Maia interacts with your local (Ollama) setup, as this might help point you in the right direction. Maia works by connecting to an OpenAI-compatible endpoint URL that you configure. Once that URL is set, all inference requests are simply forwarded to that endpoint. Maia itself has no visibility into, or control over, what happens behind that endpoint (e.g., which GPU Ollama decides to use for processing). This means that GPU selection is entirely managed by Ollama, not by Maia/Mendix Studio Pro. So the behavior you're seeing, where GPU 0 is being used instead of your NVIDIA GPU 1, is most likely something that can be resolved on the Ollama configuration side.