Skip to content
Local AI Wizard
Your machine
Theme

Local embedding models

Text · Vectors for semantic search and RAG, and rerankers.

MTEB (Multilingual, v2)

Mean(task) · 0–100 · higher is better

Source: MTEB results dataset (independent runs, CC0) · 15 curated models; other versions and benchmarks are not mixed in

MTEB

Qwen3-Embedding 8B: 70.6 (Multilingual v2, independent run)

Catalog

80 models, 13 of them rerankers

Size

28 models under 500M parameters

PT-BR

17 of 19 curated models have Portuguese confirmed by a source

Choose your machine and the charts and tables mark what fits in its memory.

80 models · catalog checking Sep 26, 2026

Embedding models
Qwen3-Embedding 8BAlibaba (Qwen) · text · PT-BR · Apache-2.070.67.6~12 GB4,096Apache 2.0
Qwen3-Embedding 4BAlibaba (Qwen) · text · PT-BR · Apache-2.069.54~7 GB2,560Apache 2.0
Harrier OSS v1 (0.6B)Microsoft · text · PT-BR · MIT69.00.6~3 GB1,024MIT
Qwen3-Embedding 0.6BAlibaba (Qwen) · text · PT-BR · Apache-2.064.30.6~3 GB1,024Apache 2.0
Multilingual E5 Large Instructintfloat · text · PT-BR · MIT63.20.56~3 GB1,024MIT
EmbeddingGemma (300M)Google · text · PT-BR · Gemma Terms of Use61.10.3~3 GB
published 200 MB
768
Matryoshka ↓128
Publisher terms: Gemma Terms of Use
BGE-M3BAAI · text · PT-BR · MIT59.60.57~3 GB1,024MIT
Nomic Embed Text v2 (MoE)Nomic AI · text · PT-BR · Apache-2.057.60.48~3 GB768
Matryoshka ↓256
Apache 2.0
Multilingual E5 Baseintfloat · text · PT-BR · MIT57.00.28~3 GB768MIT
Snowflake Arctic Embed L v2.0Snowflake · text · PT-BR · Apache-2.057.00.57~3 GB1,024
Matryoshka ↓256
Apache 2.0
Granite Embedding 311M Multilingual R2IBM · text · PT-BR · Apache-2.056.00.31~3 GB768
Matryoshka ↓128
Apache 2.0
Granite Embedding 278M MultilingualIBM · text · PT-BR · Apache-2.053.70.28~3 GB768Apache 2.0
BGE Large EN v1.5BAAI · text · PT-BR · MIT45.10.34~3 GB1,024MIT
BGE Base EN v1.5BAAI · text · PT-BR · MIT44.50.11~3 GB768MIT
BGE Small EN v1.5BAAI · text · PT-BR · MIT43.80.033~3 GB384MIT
BGE Multilingual Gemma2BAAI · text · Gemma Terms of UseNo sourced MTEB result9.2~14 GB3,584Publisher terms: Gemma Terms of Use
Granite Embedding Small English R2IBM · text · Apache-2.0No sourced MTEB result0.048~3 GB384Apache 2.0
Llama Nemotron Embed 1B v2NVIDIA · text · PT-BR · NVIDIA Open Model License Agreement (also Llama 3.2 Community Model License Agreement)No sourced MTEB result1.2~4 GB2,048
Matryoshka ↓384
Publisher terms: NVIDIA Open Model License Agreement (also Llama 3.2 Community Model License Agreement)
Nemotron 3 Embed 1BNVIDIA · text · PT-BR · OpenMDW-1.1No sourced MTEB result1.1~4 GB2,048OpenMDW 1.1
all-MiniLM-L6-v2sentence-transformers · text · apache-2.0No sourced MTEB resultNot reported, or a community distributionUnknown parameters: no estimateNot curated yetApache 2.0
all-MiniLM-L6-v2-4bitmlx-community · text · apache-2.0No sourced MTEB resultNot reported, or a community distributionUnknown parameters: no estimateNot curated yetApache 2.0
all-mpnet-base-v2sentence-transformers · text · apache-2.0No sourced MTEB result0.1~3 GBNot curated yetApache 2.0
bge-base-enBAAI · text · mitNo sourced MTEB result0.1~3 GBNot curated yetMIT
bge-base-zhBAAI · text · mitNo sourced MTEB result0.1~3 GBNot curated yetMIT
24 of 80