There is no per-unit API price: self-host only.
There is no free API. The weights are open, so you can run it yourself.
#8 of 27 in Multimodal.
On MAEB v1 audio-only · Any-to-any Retrieval · multilingual (4 languages), Qwen 2 Audio 7B scores 37.6 vs 37.6 — about the same.
Released 2024-08-09. It is the current version in its line.
Yes, the weights are open (MIT licence).
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Qwen 2 Audio 7B is an open-weights embeddings & retrieval model from Alibaba (Qwen), released Aug 2024.
Jina Embeddings v5 Omni Nano (better), Larger Clap Music And Speech (open weights).