Enterprise AI: Managing Model Evolution & Costs - Straive Report

The next phase of enterprise artificial intelligence will focus on managing continuous upgrades, costs, and performance of AI models, rather than just deploying new ones. This is according to a report by Straive. Generative AI has moved beyond experimentation, with organizations using large language models in many areas. The report states that AI helps teams complete tasks like searching and summarizing in minutes instead of hours. However, foundation models are continuously changing, presenting a new challenge. Every new model release can improve reasoning and accuracy, but it can also change pricing and output quality. Moving from one model generation to another is not like a traditional software update. Companies may need to reassess prompts and safety measures. What's interesting is the changing cost structure. Unlike traditional software licensing, foundation models use consumption-based pricing. Costs depend on factors like input and output tokens. For example, moving from Gemini 2.0 Flash to Gemini 2.5 Flash increases input token pricing from 10 cents to 30 cents per million tokens. Output token pricing rises from 40 cents to $2.50. Gemini 3.5 Flash further increases these costs to $1.50 for input and $9.00 for output tokens. Changing reasoning settings alone can create cost differences of up to four times. The report urges companies to select models based on specific business needs, not just the newest or cheapest. This means businesses need to understand and manage the ongoing evolution and cost implications of their AI systems.