Comparisons
Embedding Models in 2026: OpenAI text-embedding-3 vs BGE vs Cohere vs Voyage
Benchmark MTEB leaderboard leaders, dimensionality tradeoffs, Matryoshka embeddings, and pricing per 1M tokens.
3 min
Matryoshka Representation Learning (MRL)
OpenAI's `text-embedding-3-large` supports Matryoshka dimension shortening: truncating vectors from 3072 to 1024 dimensions cuts RAM and index storage by 66% with virtually no retrieval quality loss.