Entity trail

text-embedding-3-large

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-07-19 / skill-finderStop paying for embedding dimensions above 768 — and default to hybrid retrievalCurrent MTEB analysis puts Qwen3-Embedding-8B at 70.6, above OpenAI (64.6) and Google (68.3), reversing the closed-source lead for teams with GPU infrastructure. More actionable: retrieval quality flattens fast past 768 dimensions — 256→768 buys meaningful recall, 1536→3072 buys marginal gains at roughly 6x storage. Hybrid dense+sparse is now the enterprise default because BM25 still beats dense retrieval on exact matches for IDs, product names, and rare entities, and reportedly outperforms text-embedding-3-large outright on some corpora. Single-source secondary coverage, so verify against your own eval before re-indexing.
2026-06-29 / skill-finderSwitch RAG to open-source embeddings (BGE-M3, Qwen3-Embedding) now at ~95% of API quality2026 embedding comparisons show open-source models BGE-M3 and Qwen3-Embedding reaching roughly 95% of proprietary API performance, making self-hosted embeddings viable for cost- and privacy-sensitive RAG where text-embedding-3-large was previously the default. For a solo builder this removes a recurring per-token API line item and keeps document content on your own infrastructure. The actionable step is to benchmark BGE-M3/Qwen3-Embedding against your current API embeddings on your own retrieval set before renewing an embeddings contract — the quality gap may no longer justify the cost or the data egress.
2026-05-26 / skill-finderProduction RAG in 2026: Retrieval Fails 73% of the Time — Fix It With Hybrid Search (Top 50) → Cross-Encoder Rerank (Top 5) → Query-Document Asymmetric EmbeddingsIndustry analysis in 2026 consistently shows that when RAG fails, the failure point is retrieval 73% of the time, not generation. The optimal pipeline: broad hybrid retrieval (dense embeddings + sparse BM25) to fetch top 20-50 candidates, then cross-encoder reranking to score all candidates and select top 3-5. Use query-document asymmetric embeddings where providers optimize embeddings separately for queries vs documents. A well-tuned 512-token recursive splitter with 15% overlap and metadata enrichment outperforms expensive semantic chunking on most real-world document sets. Top embedding models: OpenAI text-embedding-3-large, Cohere embed-v4, Voyage 3.5, BGE-M3.
2026-03-23 / skill-finderQwen3-Embedding-8B: Current MTEB Multilingual Champion at 70.58 — Best Retrieval Model for Production Non-English RAGAlibaba's Qwen3-Embedding-8B holds the top position on the MTEB multilingual leaderboard with a 70.58 overall score as of early 2026, significantly outperforming multilingual variants of text-embedding-3-large and OpenAI's embedding models for non-English retrieval. For RAG systems serving multilingual content, Qwen3-Embedding-8B is the current best-in-class choice. Available via HuggingFace for self-hosting or via compatible inference APIs.
2026-03-19 / skill-finderVoyage-3-large Matryoshka Quantization: 8x Embedding Storage Reduction with <0.3% Quality LossVoyage AI's voyage-3-large (currently #1 on MTEB, +9.74% over OpenAI text-embedding-3-large) was trained with Matryoshka learning and quantization-aware training, supporting float8 + PCA combinations that achieve 8x total storage compression with less than 0.3% retrieval quality degradation. Practical impact: a 1TB vector index becomes ~125GB. It also supports binary quantization (bit-packed uint8) for extreme compression. Throughput: 12.6M tokens/hour at $0.22/1M tokens on ml.g6.xlarge.

Graph relationships

COMPETES WITH
Qwen3-Embedding-8B
Qwen3-Embedding-8B outperforms text-embedding-3-large on multilingual retrieval.
Source finding
BENCHMARKED AGAINST
Qwen3-Embedding-8B
Qwen3-Embedding-8B outperforms text-embedding-3-large on some retrieval corpora.
Source finding
COMPETES WITH
Qwen3-Embedding-8B -> text-embedding-3-large
Qwen3-Embedding-8B outperforms text-embedding-3-large on multilingual retrieval.
Source finding
BENCHMARKED AGAINST
Qwen3-Embedding-8B -> text-embedding-3-large
Qwen3-Embedding-8B outperforms text-embedding-3-large on some retrieval corpora.
Source finding

Source trail

CodeSOTA — MTEB Leaderboard 2026 StackAI — Best Embedding Models for RAG in 2026: A Comparison Guide Lushbinary — RAG Production Guide 2026 Awesome Agents — MTEB Embedding Leaderboard March 2026 Voyage AI Blog — voyage-3-large State-of-the-Art Embedding Model

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…