$0.02 per 1M tokens.
There is no free API; the cheapest paid price is $0.02 per 1M input tokens. The weights are open, so you can run it yourself.
#51 of 89 in Multimodal.
On MIEB lite · Document Understanding · English, EmbeddingGemma 2 scores 67.1 vs 73.3 — lower.
Through the Google API.
Released 2026-10-06. It is the current version in its line.
Yes, the weights are open (gemma licence).
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
EmbeddingGemma 2 is an open-weights embeddings & retrieval model from Google, released Oct 2026.
Jina Embeddings v5 Omni Nano (better), Argus Colqwen3.5 9B v0 (open weights).
BetterJina Embeddings v5 Omni NanounstatedOpen+11%same$0.02Memory (4-bit): 0.2 GB → 0.6 GBDOpenArgus Colqwen3.5 9B v0unstated+34%——Memory (4-bit): 0.2 GB → 5.3 GB
Nemotron Colembed VL 8B v2Open+33%——Memory (4-bit): 0.2 GB → 5.2 GBDArgus Colqwen3.5 4B v0unstatedOpen+33%——Memory (4-bit): 0.2 GB → 2.8 GBvVultronRetrieverPrime Qwen3.5 8BOpen+32%——Memory (4-bit): 0.2 GB → 5 GBvVultronRetrieverCore Qwen3.5 4.5BOpen+32%——Memory (4-bit): 0.2 GB → 2.7 GB