Lumenfall provider extension for llms

[mikewirth]

Hi llms!

We're Lumenfall, an AI image gateway with an OpenAI-compatible API, and have created an extension for llms.
Lumenfall has all the top AI image models (55 as of right now) and routes to multiple providers for each model, providing resiliency.

Our llms.py extension was just released, offering image generation and editing. We've tested it with the llms CLI and the Web UI.
Extension Repo
Docs

We offer an OpenAI-compatible API (/images/generations and /images/edits) and normalize all parameters, so all the model and provider quirks are abstracted. This means the aspect ratio the user chooses in llms, is translated to params the chosen model understands. (e.g. gpt-image-1.5 only takes size in the form of 1024x1024)

Quickstart

llms --add lumenfall-ai/llmspy-lumenfall
export LUMENFALL_API_KEY=lmnfl_your_api_key

Add the provider to ~/.llms/llms.json:

{
  "providers": {
    "lumenfall": { "enabled": true, "npm": "llmspy_lumenfall" }
  }
}

Then generate an image:
llms --out image "A capybara relaxing in a hot spring" -m grok-imagine-image

We'd be happy if you could try it out. New users get $1 in free API credits right now. We think the extension adds a lot of value for llms users.

If you're interested in integrating the provider more tightly into llms itself and mention it in the docs, that would be amazing!

Let us know about any bugs you find or if we should add anything.

[mythz]

Great stuff, did you want to add a fork of the extension repo in llmspy/repositories so users can discover it with llms --add?

Feel free to reply with a 1 line description in markdown of how you want it described in the docs.