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.
#186 of 200 in Embeddings.
On MTEB Russian v1.1 · Pair Classification · Russian, Arabic English Sts Matryoshka scores 68.3 vs 69.5 — lower.
Released 2024-10-13. It is the current version in its line.
Yes, the weights are open (Apache-2.0 licence).
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Arabic English Sts Matryoshka is an open-weights embeddings & retrieval model from omarelshehy, released Oct 2024.
Qwen 3 Embedding 8B (better), Giga Embeddings Instruct (open weights).
BetterQwen 3 Embedding 8BunstatedOpen+4%—$0.01Languages: not verified+1aOpenGiga Embeddings Instructunstated+16%——Memory (4-bit): 0.3 GB → 1.9 GBaBetterGiga Embeddings Instruct 3B 0826unstatedOpen+6%——Memory (4-bit): 0.3 GB → 1.9 GBaBetterGiga Embeddings Instruct 10B A1.8B 0826unstatedOpen+5%——Memory (4-bit): 0.3 GB → 6.3 GB