Today's take
Agent engineering toolchains are accelerating maturation, with CLI agents maintaining leading usage. Kimi K3.1 Pro shows a 256% usage surge, while content generation projects attract significant attention, creating positive feedback between open source and application ecosystems.
Open source
The open source landscape shows two key trends: 1) Continuous improvement in agent engineering toolchains, with morluto/rea leading with 5,901 stars in 1 day, providing foundational capabilities for agent development; 2) Significant attention for content generation projects, as seen with hugohe3/ppt-master's 805 stars increase placing it in the top 14, demonstrating AI's engineering capabilities in presentation creation. #1 morluto/rea's 382,532 stars and #3 Zagitta/gpu-unlock's 4,888 stars increase represent progress in agent toolchains and hardware optimization respectively.
Models
Top 5 models show overall usage declines of 6-35%, but emerging models like Kimi K3.1 Pro show explosive growth. Kimi K3.1 Pro jumps to 12th place with 291B tokens, a 256% increase from the previous day, reflecting enhanced competitiveness of domestic models in specific scenarios. Multiple models like Nemotron 3.5 Super VL (free) show over 20% usage growth, indicating increased adoption of multimodal capabilities in developer communities.
Apps
CLI agents continue to dominate the application ecosystem, with Claude Code maintaining first place with 1.58T tokens. Emerging apps like pi and KoalaBear show impressive performance in the rising category, with pi achieving 746B tokens and 5.6M requests, demonstrating the market potential of personalized AI tools. The application ecosystem shows synergy with open source projects, as seen in the correlation between #16 affaan-m/ECC's toolset and CLI agent usage growth.
What to watch
The two trends of agent engineering and content generation will continue to deepen. Pay attention to the interplay between model usage growth and open source project activity, especially the practical implementation of emerging models in specific application scenarios.
