Jan: Ollama integration for existing model support

[eckartal]

Overview

• People store models they downloaded from Ollama. In Jan, we should let them run models they already downloaded. This is probably the most commonly asked feature from r/LocalLLaMA.

Goals

• Implement a feature to detect and import Ollama models into Jan

A note

• Plus, we can solve this at Cortex-level: https://github.com/orgs/janhq/discussions/3622
• Cortex is a long-term solution, for quick solutions we should consider Ollama support

[harlley]

just added a new provider and it worked

[TomLucidor]

This needs to be an FAQ thing, but does it work with MLX?

[jade-city2]

It doesn't support thinking or context length or anything like that.

[ringe]

I can't get Ollama to work

[GIN] 2026/05/27 - 09:03:50 | 200 |     703.303µs |       127.0.0.1 | GET      "/v1/models"
[GIN] 2026/05/27 - 09:05:31 | 403 |      29.996µs |       127.0.0.1 | POST     "/v1/chat/completions"
[GIN] 2026/05/27 - 09:05:31 | 403 |      11.311µs |       127.0.0.1 | POST     "/v1/chat/completions"

Is there some particular setting in Ollama required?

[ringe]

CORS problem in the Ollama setup: ollama/ollama#4115 (comment)