Mealissa routes any model in the Ollama library — Llama, Qwen, Gemma, Mistral, Phi, DeepSeek, and more. There's no special list to maintain: pull a model with Ollama and it's instantly reachable through your Mealissa endpoint.
Meta's Llama family — versatile general-purpose models for chat, reasoning, and code.
Alibaba's Qwen models, strong at multilingual tasks, reasoning, and tool use.
Google's lightweight Gemma models — efficient and capable on modest hardware.
Fast, compact Mistral and Mixtral models with excellent quality-to-size ratios.
Microsoft's Phi small language models — punchy reasoning in a tiny footprint.
DeepSeek models for deep reasoning, coding, and long-context workloads.
…and every other model in the Ollama library. Whatever you can run locally, Mealissa makes globally accessible.