With the summer holidays over, we want to welcome you back with two new models available on Nebula: GPT-OSS:20B and Gemma3:12B. Along with the new models we also want to present some of our favourite features of Nebula.

• OpenAI API compatibility: The Nebula API is available through the standard OpenAI API access (the openai python package).
• Knowledge: Workspace → Knowledge allows you to upload documents which the LLMs can use to give you more relevant responses. You can reference the uploaded document(s) or Collections using ‘#‘ at the beginning of the prompt and selecting the desired document(s). You can try out this feature by using #Nebula and asking the LLMs information about Nebula. For more information see https://docs.openwebui.com/features/workspace/knowledge. Note: this feature also works with links to websites, for example #https://docs.openwebui.com/features/workspace/knowledge
• Transcript generator: Sound files can be uploaded to Nebula, and users can ask the LLMs to generate the transcript of the recording (tested with GPT-OSS).
• Server expansion tested: Last week we tested the option to expand Nebula, and the test was a success, both in terms of response time, and concurrent use. This will allow us to expand the capacity easily in the future.

We are currently transitioning to a model where researchers can support Nebula. The prices here are NOT set in stone, and feedback is more than welcome and appreciated.

Please send feedback about the costs to r.apsan@vu.nl and m.otte@vu.nl.