There is no per-unit API price: API price not public.
There is no free API. The weights are open, so you can run it yourself.
#107 of 200 in Embeddings.
On MTEB Indic v1 · STS · multilingual (13 languages), Bekko Embedding v1 A25m scores 49.6 vs 50.4 — about the same.
Released 2026-07-28. 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.
Bekko Embedding v1 A25m is an open-weights embeddings & retrieval model from hotchpotch, released Jul 2026.
Qwen 3 Embedding 8B (better), STS Multilingual Mpnet Base v2 (open weights).
BetterQwen 3 Embedding 8BunstatedOpen+24%—$0.01Memory (4-bit): 0.1 GB → 4.5 GBGOpenSTS Multilingual Mpnet Base v2unstated+71%——Max input: 8K tokens → 0K tokens+1iBetterMultilingual E5 Large InstructunstatedOpen+8%—$0.01Max input: 8K tokens → 1K tokens+1
BetterBge M3unstatedOpen+5%—$0.01Memory (4-bit): 0.1 GB → 0.3 GB
Text Multilingual Embedding 002unstated+8%—$0.025Max input: 8K tokens → 2K tokens
KaLM Embedding Gemma3 12B 2511Open+28%——Memory (4-bit): 0.1 GB → 7.1 GBOOcten Embedding 8BOpen+25%——Memory (4-bit): 0.1 GB → 4.5 GBLQwen Indic v1Open+24%——Memory (4-bit): 0.1 GB → 4.5 GB