Plotting AI model release cadence: two labs are accelerating, three aren't (swiftalerts.trade)

Recent analysis by Ethan Mollick examines the release cadence of AI models from various labs, indicating a significant divergence between labs that are accelerating their releases and those that are not. Anthropic and OpenAI are showing marked increases in their model release rates, suggesting that both may be experiencing a feedback loop of self-improvement where their latest models enhance the development of subsequent versions. In contrast, other labs like Google and Meta have not shown similar acceleration, with Meta's releases plateauing after April 2025. This observation holds implications for the AI/ML community, as it raises questions about the future competitive landscape. If the pattern continues, the gap between the accelerating labs and those stagnating could widen, reinforcing the notion that self-improvement mechanisms are driving faster advancement. Key technical factors include the potential impact of enhanced compute efficiency—exemplified by Tri Dao's recent advancements in GPU utilization—and increased talent flow to faster-releasing labs, which further support the recursive self-improvement argument. This trend could critically influence compute resource investment decisions and overall market dynamics in the AI sector.