Perplexity, a cutting-edge AI startup, just launched an API that uses two online large language models (LLMs), pplx-7b-online and pplx-70b-online, to help users get real-time information. Unlike traditional offline LLMs, these models pull data directly from the internet, making them better at answering questions quickly with the latest information.
The key feature of these models is their use of an API, which is unique in the world of AI. Other online LLMs are available, but Perplexity’s models are the first to offer this functionality to developers via an API. This means that the models provide real-time “snippets” of information from high-quality websites.
What sets Perplexity’s LLMs apart is that the company has fine-tuned them for optimal performance and integrated them with advanced technology. To ensure the models deliver accurate, up-to-date, and factual information, Perplexity relied on top-tier training sets.
The results of these evaluations reveal that pplx-7b-online and pplx-70b-online consistently outperformed their counterparts in freshness, factuality, and overall preference. For instance, in the freshness criterion, pplx-7b and pplx-70b achieved estimated Elo scores of 1100.6 and 1099.6, surpassing gpt-3.5 and llama2-70b. Developers can now access Perplexity’s API to create applications using these models, and there are special plans available for early testers.
This pioneering release by Perplexity introduces a paradigm shift in AI-driven information retrieval systems, showing superior performance in delivering accurate, up-to-date, and factual information.