UBC NLP

GTR-XL

Strong open-source embeddings with focus on retrieval.

High retrieval performanceOpen-sourceGood recall on various datasetsSupports longer contexts
Today's score
87.0
Try GTR-XL

Where it ranks today

Best for / Not great for

Best for
  • Document retrieval
  • RAG systems
  • Semantic search
  • Information extraction
Not great for
  • Zero-shot classification
  • Highly specialized domains without fine-tuning
  • Very low latency requirements
  • Applications needing multilingual support out-of-the-box

Why it ranks here

While newer models have emerged, GTR-XL remains a solid choice for retrieval-focused tasks, offering strong open-source performance. Its effectiveness in finding relevant documents makes it suitable for RAG and search engines, though it sees less buzz than cutting-edge models.

30-day trend

Score breakdown

Search trends90
Benchmarks87
Developer buzz88
News mentions85

Pricing

API: $0.00 in · $0.00 out per 1M tokens · Consumer: $0.00/mo

Pricing plans

Popular
Self-Hosted
Deploy and customize your own retrieval model.
Free
  • Open-source
  • High recall
  • Customizable
  • Requires self-hosting
Download model
Managed Inference
Access GTR-XL through API providers.
$0 /usage
  • Scalable
  • API access
  • Usage-based billing
Find providers
Compare with another modelHow is this score calculated? →Snapshot 2026-09-05