🤗Proud to see Sentence Transformers' all-MiniLM-L6-v2 trending on @huggingface! Now stop using it, it's not 2021 anymore. You're leaving performance on the table.

Sep 16, 2026 · 8:06 AM UTC

10
9
97
4,884
Have you tried this one? huggingface.co/ibm-granite/g…

ibm-granite/granite-embedding-30m-english · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co
1
1
8
446
Or maybe this one? huggingface.co/Snowflake/sno…

Snowflake/snowflake-arctic-embed-s · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co
1
6
354
Or: look at all Sentence Transformers models and pick something new: huggingface.co/models?librar… There's a ton of options.

Models compatible with the sentence-transformers library – Hugging Face

Explore machine learning models.

huggingface.co
1
4
291
Not sure what to choose? Consider looking at the MTEB leaderboard as a start, although I think you're best off evaluating on your own data for a final pick: huggingface.co/spaces/mteb/l…

MTEB Leaderboard - a Hugging Face Space by mteb

Embedding Leaderboard

huggingface.co
6
273
Sort replies: Relevant Recent Liked
Replying to @tomaarsen @huggingface
Maybe it use for distance calculating to AI slop 😅 it should be pre-ChatGPT mostly.
1
2
652
Definitely pre-ChatGPT, it was trained in August 2021, over a year before ChatGPT
3
228
Replying to @tomaarsen @huggingface
They should induct this one in the hall of fame On that note HF should def have a hall of fame for legendary models like this one lol
1
1
82
Great idea 👀 We've got to get bert-base-uncased and gpt2 in there
77
Replying to @tomaarsen @huggingface
The benchmarks on your own leaderboard puts minilm above the two models you suggested...
1
1
116
Oh, that surprises me. Do you mean MTEB? On the English leaderboard, all-MiniLM-L6-v2 isn't great. Snowflake beats it, and R2-30m isn't on there it seems.
1
1
102
more replies
Replying to @tomaarsen @huggingface
I just tend to look for modernbert based ones
2
67
Replying to @tomaarsen @huggingface
It is a solid model and aged like fine wine 🍷
2
78
Replying to @tomaarsen @huggingface
one of the classics
17
Replying to @tomaarsen @huggingface
@JinaAI_ models ftw!!!
2
33
Replying to @tomaarsen @huggingface
that model is mini and mighty. If properly fine-tuned can beat newer and bigger ones. A very strong baseline.
1
36
Replying to @tomaarsen @huggingface
static-retrieval-mrl-en-v1 is the drop-in for anyone who stayed on MiniLM for speed: same API, and Matryoshka lets you trim dims.
72