OpenAI's text-embedding-3-large model has dropped to 13th place out of 15 on the 2026 aggregate embedding leaderboard, scoring 58.96. This performance is significantly lower than the top-ranked Qwen3-Embedding-8B, which achieved a score of 70.58 and is available under an Apache 2.0 license at no cost. Even smaller variants of Qwen3-Embedding can outperform OpenAI's model, highlighting a potential issue in the retrieval layer of many Retrieval-Augmented Generation (RAG) systems that often rely on outdated embedding endpoints. AI
IMPACT Highlights potential performance issues in RAG systems and suggests a shift towards more efficient, open-source embedding models.
RANK_REASON Article analyzes performance of existing models and provides an opinion on switching, rather than announcing a new release or significant industry event.
- 2026 aggregate embedding leaderboard
- Apache Software License 2.0
- GPT-5.6
- Grok
- LangChain
- MacBook
- OpenAI
- Opus 4.8
- Qwen3-Embedding-8B
- retrieval-augmented generation
- text-embedding-3-large
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
