Large Language Models¶
🤖 Embeddings¶
CEM-014-000
Specification¶
If embedding data and accuracy is very important, use Voyage-3.5
Motivation¶
Model |
Provider |
Strengths |
Dimensions |
Latency |
Best For |
|---|---|---|---|---|---|
Voyage-3.5 |
Voyage AI |
SOTA for RAG; context-aware |
1024 (Flex) |
Low (API) |
High-precision QA |
Embed-v4.0 |
Cohere |
Multilingual & Multi-modal |
1024 |
Low (API) |
Global, diverse data |
text-embed-3 |
OpenAI |
Massive adoption, stable API |
1536/3072 |
Medium (API) |
General purpose |
BGE-M3 |
BAAI (OSS) |
Multi-lingual & Sparse/Dense |
1024 |
Var (Host) |
Private/On-prem |
Qwen3-Embed |
Alibaba (OSS) |
Instruction-aware, efficient |
512–1024 |
Ultra Low |
Small-data agility |