V
Vijay Bharadwaj
Hi. Currently Llama 4 Scout and Maverick are available through API for use. Internally however, we've found these models to be of subpar quality, and we have issues adding them to our pipeline because they've redone the architecture, which is causing compatibility issues with our internal pipelines.
If we go ahead and release it on Merlin today, you will not only use an inferior model for the cost, but also a model that won't go very well with Merlin's AI features like Crafts, Chat with Image, etc.
So, please decide if you REALLY want it -- then we'll probably move forward with adding it.
P.S. Llama 4 adds no new value to the model list that the other models can't do already, at the same or lower query cost.
monde mamon
Vijay Bharadwaj you're right, it really is useless to implement, most would only use it for the novelty
J
Joey
Vijay Bharadwaj Yeah exactly, I’d really only use it out of curiosity myself. Maybe it could be an idea to have some kind of “advanced mode” within Merlin where features like this can be enabled. Most users probably don’t need it, but for those who want to experiment a bit more or dive deeper, it could be interesting.
V
Vijay Bharadwaj
Merged in a post:
Add Llama 4.0
D
Digital Dexter
The older models need to be upgraded, or the new models added.
P.S. Adding new models at a reasonable credit cost is important, more than other developments that you are doing right now.
V
Vijay Bharadwaj
Merged in a post:
Please add LLAMA 4
J
Joey
Could the new AI-models based on LLAMA4 (Scout, Maverick and Behemoth) be added as soon as the API is available?