Rather than relying on benchmarks, this talk builds on our State of AI report released in December 2025 — one of the largest empirical studies of real-world LLM usage to date, analyzing 100 trillion tokens of inference data from the OpenRouter platform.

We examine how models are actually used in practice across tasks, geographies, and time.

Among the original findings:

• Open-weight models now serve roughly a third of all tokens in an increasingly fragmented ecosystem
• Creative roleplay is the dominant open-source use case
• Programming is the fastest-growing category overall

We’ll revisit these findings with fresh data and updated trends, including the continued rise of agentic inference (multi-step, tool-integrated workflows) alongside cost-usage dynamics, geographic shifts, and model retention patterns.

We’ll also discuss what these evolving patterns mean for model builders, infrastructure operators, and investors.