Back to Arena

Mixedbread AI

by Mixedbread AI

System Card

OrganizationMixedbread AI
Released2024-01
Architectureexternal-memory-network / embedding and reranking model API
DetailsMixedbread provides mxbai-embed-large (Matryoshka embedding model supporting dimensionality reduction) and mxbai-rerank (GRPO-trained rerankers supporting 100+ languages) via API and open-source HuggingFace weights. The 2D-Matryoshka technique allows simultaneous reduction of layers and dimensions. The Mixedbread Platform alpha integrates embeddings, reranking, document parsing, and vector stores.
Parameters—
Domainrag-retrieval
Open SourcePartial
WebsiteVisit
Matryoshkarerankingopen-weightsmultilingualGRPO

Capability Profile

Benchmark Scores

5 of 14 benchmarks
Long-Context Retrieval
2/5
RULER
71.956p
NIAH
no data
LooGLE
no data
LongBench
603p
∞Bench
no data
Multi-Turn Recall
0/2
LoCoMo
no data
MemoryBank
no data
Cross-Session Memory
0/1
LongMemEval
no data
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
72.556p
HotpotQA
68.440p
Agent Task Memory
0/1
AgentBench-Mem
no data
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
1/1
RAGAS
67.537p
Sources:Mixedbread AI vendor documentation; evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Mixedbread AI vendor documentation; evaluated on LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (Tsinghua KEG, 2308)Mixedbread AI vendor documentation; evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)Mixedbread AI vendor documentation; evaluated on RAGAS: Automated Evaluation of Retrieval-Augmented Generation (Exploding Gradients, 2309)Mixedbread AI vendor documentation; evaluated on RULER: What's the Real Context Size of Your Long-Context Language Models (NVIDIA, 2404)