Nvidia Launches New Top Performing AI Embedding Models
Nvidia released three new open models designed to help AI understand and retrieve information more accurately. The largest version, an 8 billion parameter model, is now the highest ranked system on the popular MTEB benchmark. These tools are built to handle very long documents with up to 32,000 tokens.
The smaller 1 billion parameter versions are designed for efficiency. Nvidia used advanced techniques to compress these models, allowing them to keep nearly all their accuracy while running much faster. This makes them a great choice for companies that need quick search results.
You can now use these models to build smarter search and data retrieval systems. The new versions work especially well on Nvidia hardware, offering better performance for your computing budget. These models are available today for developers who want to improve their AI applications.
Comments (0)
No comments yet. Be the first!
More AI news
NewsNetflix is using AI in 300 projects
Netflix is quickly integrating artificial intelligence into its film and television production process to save time and money.
NewsMoonshot AI Launches Kimi K3 Model
Moonshot AI has released Kimi K3, a powerful new model designed to handle massive amounts of data.
NewsRun Powerful AI on Your Phone with Bonsai 27B
PrismML has released a new, lightweight version of the Qwen 27B model that fits on everyday devices.