OpenAI remains the top choice among AI developers, particularly in large language models. According to TechRadar, citing a Vercel study, 87% of the 656 developers surveyed use OpenAI models, and 83% work with its API. However, developers increasingly diversify: most now rely on at least two providers, and 60% have switched in the past six months.
This growing trend toward exploration has seen nearly one-third of enterprise users adopt DeepSeek. In contrast, only 10% opt for Microsoft’s language models. Although OpenAI still dominates, the presence of multiple model providers is becoming standard practice. We’ll keep you updated as these preferences continue to evolve.
Practical Use and Lack of Dedicated AI Teams
AI product development today centers around specific, clearly defined use cases. Developers and companies are focused more on available tools than on in-house expertise. In fact, 45% of companies do not have a dedicated AI department, and 57% report no AI experts at the leadership level.
Customer-facing applications are a priority area, with 75% including AI-related features. Chatbots, for example, are now used in 39% of support services. This emphasis on functionality over internal teams reflects a pragmatic approach to integrating AI, especially amid rapid changes in the technology landscape.
Managing Costs Through Smart Tools
Cost remains a significant concern. More than 70% of developers test AI models manually and keep monthly expenses under $1,000. Training models is rare—only 14% do so—while the majority rely on Retrieval-Augmented Generation (RAG) and vector databases to reduce costs without compromising functionality.
Despite the hype surrounding AI, respondents gave the overall trend an average hype rating of 6.4 out of 10. Still, when asked about AI’s potential to transform their industry within the next year, developers offered a more optimistic rating of 7.7.
This contrast shows cautious optimism. While the industry buzz remains high, most developers are looking for grounded, cost-effective applications. And as alternatives to OpenAI gain traction, we’ll keep you updated on how developer preferences continue to adapt.