UAE-Large-V1
AnglE-optimized embeddings — MIT-licensed with a novel angle-based training objective.
About
UAE-Large-V1 (late 2023, WhereIsAI) introduced AnglE (Angle-optimized Embeddings): training in angle space rather than cosine similarity, which the authors showed improves retrieval quality. 760k Hugging Face downloads.
MIT licensed — commercial use fully allowed. The methodological novelty (AnglE) earned it real attention and strong MTEB results at release. A 2023 335M model now superseded by BGE-M3 and MXBAI on the quality-efficiency frontier, but the AnglE objective influenced subsequent work. For researchers studying embedding objectives; for deployment, BGE-M3 is the current answer.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-claimed at release (2023): strong MTEB via AnglE — treat as provider-claimed and dated. Methodological influence is real.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier). The AnglE objective was a genuine methodological contribution with strong release-time results, MIT-licensed, and solid adoption (760k downloads). Docked three points: (1) 2023-era, superseded by BGE-M3/MXBAI; (2) 512-token context; (3) the WhereIsAI project has limited follow-through. For embedding-objective researchers.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 4.0/10
- Openness · license + docs
- 8.8/10
- Manual review · sanity read + red-flag check
- 6.0/10
- · HF downloads 759,810 (peer-normalized)
- · released 1094 days ago
- · manual review: AnglE objective was a genuine contribution; 2023-era, superseded; limited follow-through; license: MIT