FeaturedRetrieval Augmented Generation
Think building your own Retrieval-Augmented Generation (RAG) system will give you a competitive edge? Think again. Companies are sinking months and hundreds of thousands of dollars into in-house AI projects, only to end up with sluggish, unsecure, and overpriced systems that can't hold a candle to existing solutions.
Read more
HCMBench is Vectara’s open-source evaluation toolkit designed to rigorously test and compare hallucination correction models. With modular pipelines, diverse datasets, and multi-level evaluation metrics, it gives developers a powerful, standardized way to measure and improve the accuracy of RAG system outputs.
AI hallucinations create significant business risks and erode user trust. Vectara's Hallucination Corrector (VHC) identifies inaccuracies, suggests fixes, and provides essential guardrails for your AI applications.
We’ve created a trusted platform for building safe and reliable AI applications and continue to invest in features that improve reliability, accuracy, and flexibility. Today, we're taking another step forward by introducing two powerful new capabilities: the Hallucination Evaluation Model (HHEM) and OpenAI Chat Completions endpoints.
Introducing Mockingbird 2: our latest grounded generation model optimized for RAG with advanced crosslingual support and improved performance. It runs securely in any environment—SaaS, cloud, or on-prem—delivering high-accuracy responses without data leakage risk.
We’re excited to announce custom text encoders, empowering users to integrate their preferred text embedding models directly into Vectara. This enhancement offers greater flexibility and control over the text processing pipeline, accommodating various project requirements.
Now you can integrate the language model of your choice into Vectara while keeping existing security approvals, fine-tuning efforts, and AI investments intact. Use Vectara’s query and chat APIs while maintaining control over your model choices.
Vectara's Intelligent Query Rewriting (Tech Preview) improves RAG search accuracy by separating queries into their natural language and filter expression components.
Vectara’s Hughes Hallucination Evaluation Model (HHEM) now supports 8 languages, expanding from English, German, and French to include Portuguese, Spanish, Arabic, Chinese - Simplified, and Korean.
Vectara's new query observability feature provides detailed insights into query history, configurations, and system behavior, empowering users to optimize search performance and build trust in AI systems.
Connect our community channel.
Join our discussion channel.
Get news, company information.
Adopt best practices in projects.
Suggest your own ideas.
Ask your follow-up questions.
