Model Terminal

cross-encoder/ms-marco-MiniLM-L6-v2

A compact cross-encoder reranking model fine-tuned on MS MARCO Passage Ranking, designed to score query–passage pairs for relevance in second-stage retrieval pipelines. It balances speed and accuracy among the MiniLM reranker family, achieving 74.30 NDCG@10 at roughly 1,800 docs/sec on a V100 GPU. Source: supabase.

Value score
Not scored
Context
tokens
Max output
tokens
Price

Capability radar

Identity

Developer
cross-encoder
Openness
Modalities
Release
Knowledge cutoff
Deprecation
API docs

Benchmarks

No benchmark scores yet.

Compare nearby

Pricing

Input / 1M
Output / 1M
Speed