New Relic has published its first AI Unwrapped: 2025 AI Impact Report, highlighting how developer choices are shaping the evolution of artificial intelligence. The findings, based on anonymised data from 85,000 active customers, show that OpenAI’s ChatGPT continues to dominate among large language models (LLMs), while developers are increasingly exploring alternative models.
ChatGPT leads but model diversity grows
The report found that ChatGPT models accounted for over 86% of all LLM tokens processed by New Relic users over the past year. Adoption has surged for newer models, particularly ChatGPT-4o and ChatGPT-4o mini, which have seen rapid uptake since their launch. The shift from ChatGPT-3.5 Turbo to ChatGPT-4.1 mini in April illustrates how quickly developers adopt newer models in search of better performance and lower costs.
Despite ChatGPT’s strong lead, New Relic observed a 92% increase in the number of unique models used in AI applications during the first quarter of 2025. This suggests growing interest in open-source alternatives, domain-specific tools, and task-focused solutions. Meta’s Llama model ranked second in LLM token usage, signalling its rising popularity among developers.
“AI is rapidly moving from innovation labs and pilot programmes into the core of business operations,” said Nic Benders, Chief Technical Strategist at New Relic. “The data from our 2025 AI Impact Report shows that while ChatGPT is the undisputed dominant model, developers are also moving at the ‘speed of AI,’ and rapidly testing the waters with the latest models as soon as they come out.”
Monitoring tools become essential for AI reliability
As AI adoption grows, enterprises are placing greater importance on monitoring solutions that help manage accuracy, compliance, and cost efficiency. New Relic’s AI Monitoring tool, launched in 2024, has seen steady growth with a 30% quarter-over-quarter increase in usage over the past year. The solution is designed to be easy to implement and use, offering value to a broad range of users from DevOps engineers to senior executives.
“This underscores that as AI is ingrained in their businesses, our customers are realising they need to ensure model reliability, accuracy, compliance, and cost efficiency,” Benders added.
Python remains top choice, with Java on the rise
Python continues to be the most widely used language for AI development due to its broad community support and extensive tooling. According to New Relic, Python usage grew by nearly 45% since the previous quarter. Node.js follows closely in both request volume and customer adoption.
However, Java has emerged as a fast-growing option, with usage up by 34% since last quarter. This trend indicates that large enterprises may be increasingly deploying production-level AI applications built in Java, particularly for LLM use cases.
The report offers insight into how rapidly evolving model performance, developer preferences, and enterprise requirements are reshaping the AI landscape. As the technology becomes more embedded in business functions, tools for monitoring and managing AI systems are becoming just as critical as the models themselves.
