Developers who use the API of answers can access the same models that light the search for chatgpt: GPT-4O search and GPT-4o Mini search. These models can Explore the web To answer questions and quote sources in your answers.
That is remarkable because OpenAi says that the additional web search capacity improves the objective precision of its AI models. In Openai’s Simpleqa benchmarkwhose objective is to measure the challenge rate, the GPT-4O search obtained 90 percent, while GPT-4o Mini Search achieved 88 percent, so it substantially exceeds the GPT-4.5 model largest without search, which obtained 63 percent.
Despite these improvements, technology still has significant limitations. In addition to problems with the Navigation websites of Cua properly, the improved search capacity does not completely solve the problem of AI conslegations, since the GPT-4O search continues to make errors in fact 10 percent of the time.
Together with the API of Answers, Openai launched the open source SDK agentsProvide free tools for developers to integrate models with internal systems, implement safeguards and monitor agent activities. This tool kit follows the previous openai launch of SwarmA frame to orchestrate multiple agents.
These are still the first days in the field of AI agent, and things will probably improve rapidly. However, at this time, the movement of the AI agent remains vulnerable to unrealistic statements, as demonstrated earlier this week when discovered users that the Chinese startup butterfly effect Manus ai Agent Platform failed to comply with many of his promises, highlighting the persistent gap between promotional statements and practical functionality in this category of emerging technology.
#Operai #pushes #agent #capabilities #developer #API
