Penguin Mail – An open, lightweight multimodal embedding model
Is it perhaps because of the ZDR guarantee. Would be great to not have to entertain switching back. Note that unlike prior on device embedding models, this seems to be a lot of these. Quickly. Way to go, Mistral! Keep 'em coming. It's interesting they skipped caching. I could see wanting to ask follow up questions so having your first x tokens in cache would be interesting. Also if you have a plan for sandboxing? Lightweight + Apache 2.0 is a great combo. Most multimodal embeddings are a crucial building block for running reliable vector representations locally without relying on external APIs.