As Nebula is built upon Ollama (link) Nebula can theoretically host all the AI models that Ollama can handle.
Models currently supported by Nebula [table with column (via SURF)]
Model |
|---|
| Gemma 3 12B |
| GPT-oss 20B |
| DeepSeek-r1 1.5B |
| DeepSeek-r1 8B |
| Llama 3.1 8B |
| Llama 3.1 70B |
| Llama 3.2-vision 11B |
| Llava 7B |
Open Source licenses used by models currently (Sept 2025) supported by Ollama.
License Name |
Description |
Example Models |
|---|---|---|
| Apache License Version 2.0 | Permissive open-source license | mistral, neural-chat, e5-base, sam, sentence-transformers |
| MIT License | Permissive open-source license | phi, phi-3-mini, whisper, clip, blip, bark, gfpgan, clipseg |
| Meta Llama 2 Community License | Meta’s custom license for Llama 2 and CodeLlama models | llama2, codellama |
| Meta Llama 3 Community License | Meta’s custom license for Llama 3 models | llama3 |
| Gemma License | Google’s custom, community license for Gemma models | gemma, codegemma |
| Dbrx Community License | Databricks’ custom license for DBRX | dbrx |
| DeepSeek License | DeepSeek’s custom license | deepseek-llm, deepseek-coder, deepseek-vl |
| Nomic AI License | Nomic AI’s custom license | nomic-embed, nomic-embed-text |
| CreativeML Open RAIL-M | Responsible AI License, model weights only (used by Stability AI) | stable-diffusion, sdxl, sdxl-turbo, controlnet |
| LLAVA License | LLAVA custom license | llava, bakllava, llava-phi, llava-yi |
| Playground License | Playground AI custom license | playground-v2, playground-v2.5 |
| BSD 3-Clause License | Permissive open-source license | real-esrgan |
