When will Embedding Models be available in Groq?

Feature Requests
1

Hi,

When will Groq add embedding models?

Thank you!

2

Hi @enzo,

We’re considering adding a couple of models but we haven’t settled on which ones to add yet.

Are there any open source ones in particular that you find useful or you’d like us to add?

1 Like
3

Hi,

Interested in multilingual embedding models with high retrieval scores (average >60 on MTEB multilingual retrieval) for mRAGs etc, ex:

BGE-M3 (BAAI/bge-m3 = 65.2 compared to Gemini-Embedding-001 = 68.32, pretty good for an open source model)

Others:
intfloat/multilingual-e5-large (64.8)
nomic-ai/nomic-embed-text-v2 (63.5)
Alibaba-NLP/gte-multilingual-base (62.9)
Qwen/Qwen3-Embedding-0.6B (62.1)

Regards
Enzo

1 Like
4

Thank you, that list if very very helpful!

5

Hi @yawnxyz

Please include mxbai-embed-large-v1 - it is in super wide usage and our default.

Regards

1 Like
6

hi @yawnxyz

we use Qwen/Qwen3-Embedding-8B, which is currently the top open source model in the MTEB leaderboard

1 Like
7

Good to know, thank you! I’m curious, do you run these locally or on something like HF Inference endpoints?

8

we currently use DeepInfra for this specific task, here is their offering:

Qwen/Qwen3-Embedding-8B - Demo - DeepInfra

1 Like
9

I like nomic-embed-text model

10

I’m curious — what are people liking about something like nomic vs. mixedbread vs. qwen and other ones?

I’ve liked using mixedbread but mostly because it’s lightweight and works fairly well; I haven’t really benchmarked or even vibe checked / cross-compared them… what makes one of these embedding models much better than another?

Are some of these better for some use cases vs. other uses cases?

11

Any roadmap/timeplan for deploying embedding models in groq production?

1 Like
Groq Java SDK [Unofficial]
12

Not yet, sorry. Embedding models are lower priority right now though

13

multilingual-e5-large would be nice, even it is an older model

1 Like