BAAI
A Chinese non-profit artificial intelligence research institution focused on foundation model research and open-source releases.
Timeline (last 90 days)
bge-small is the small-size version in the BGE series of sentence embedding models released by BAAI, with roughly tens of millions of parameters, and outputs 384-dimensional vectors.
BAAI's AREX-2 is a 27B-parameter model, released under the Apache 2.0 license and not gated, and the architecture class declared in its config.json is Qwen3_5ForConditionalGeneration.
There is currently no NVFP4 build of AREX-2; available quantized versions include BAAI's original BF16 (54.7 GB), prithivMLmods' FP8 (35.9 GB), cyankiwi's AWQ INT4 (20.2 GB), and numsu's INT4 W4A16 (18.6 GB).
BAAI/bge-small-en-v1.5 is an English general-purpose embedding model released by BAAI, with 384 dimensions, based on the BERT architecture, and a maximum sequence length of 512 tokens.
Commonly used open-source reranking models include the BGE-Reranker series and models released by BAAI.
BGE-M3 was developed by the Beijing Academy of Artificial Intelligence (BAAI); the three M's in its name stand for Multi-Linguality, Multi-Functionality, and Multi-Granularity, and it supports more than 100 languages.
All Facts (6)
BGE-M3 was developed by the Beijing Academy of Artificial Intelligence (BAAI); the three M's in its name stand for Multi-Linguality, Multi-Functionality, and Multi-Granularity, and it supports more than 100 languages.
60%Unverifiedbge-small is the small-size version in the BGE series of sentence embedding models released by BAAI, with roughly tens of millions of parameters, and outputs 384-dimensional vectors.
50%UnverifiedBAAI's AREX-2 is a 27B-parameter model, released under the Apache 2.0 license and not gated, and the architecture class declared in its config.json is Qwen3_5ForConditionalGeneration.
50%UnverifiedThere is currently no NVFP4 build of AREX-2; available quantized versions include BAAI's original BF16 (54.7 GB), prithivMLmods' FP8 (35.9 GB), cyankiwi's AWQ INT4 (20.2 GB), and numsu's INT4 W4A16 (18.6 GB).
50%UnverifiedBAAI/bge-small-en-v1.5 is an English general-purpose embedding model released by BAAI, with 384 dimensions, based on the BERT architecture, and a maximum sequence length of 512 tokens.
50%UnverifiedCommonly used open-source reranking models include the BGE-Reranker series and models released by BAAI.
50%