Today's AI Research: Fastest-Growing Projects — May 10, 2026
Today's AI Research, we're seeing a surge in innovative projects that push the boundaries of language models, multimodal intelligence, and autonomous agents. Researchers are exploring new techniques to improve the consistency and efficiency of large language models, while others are developing benchmarks and frameworks for evaluating and training more advanced AI systems.
The lukiIabs/trading-agents repository is gaining traction with a growth score of 53.94 and 233 stars, as it provides a comprehensive platform for trading agents using multi-agent finance, stocks, crypto, fintech, quantitative algo trading, and sentiment analysis. Its popularity stems from its unique approach to combining OpenAI's LLM capabilities with Node.js and JavaScript, making it an attractive solution for researchers and developers alike.
In contrast, fkyah3/opencode-yg is a research fork of opencode that demonstrates Language Anchoring, enabling LLMs to think consistently in a specific language. With 100 commits over the past 30 days and a growth score of 17.92, this project is gaining attention for its potential to improve the compliance and consistency of AI systems, particularly in languages like Chinese.
matrix-agent/awesome-agentic-world-modeling boasts an impressive 194 stars and a growth score of 8.78, despite having only 12 commits over the past month. This curated list provides a comprehensive overview of agentic world modeling, covering foundations, capabilities, laws, and beyond, making it a valuable resource for researchers in this field.
AutoMedBench/AutoMedBench is another notable project, with a growth score of 8.44 and 25 stars. As a medical auto-research benchmark for autonomous AI agents, it offers a standardized framework for evaluating the performance of AI systems in medical applications, which explains its growing popularity among researchers in this domain.
thunlp/OPD has garnered significant attention with 290 stars and a growth score of 7.43, thanks to its official repository for the paper "Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe." This project offers valuable insights into the distillation process of LLMs, making it an essential resource for researchers in this area.
Other notable projects include XIAO4579/PRISM, which explores pre-alignment via black-box on-policy distillation for multimodal RL, and gameworld-project/gameworld, a standardized evaluation framework for multimodal game agents. Both projects have demonstrated moderate growth, with PRISM scoring 5.84 and gameworld-project/gameworld scoring 5.12.
Lastly, Hedlen/Awesome-Multimodal-Intelligence and AMAP-ML/DCW are two more projects worth mentioning, despite their relatively lower growth scores of 3.68 and 3.54, respectively. The former provides a curated collection of papers, code, and datasets for multimodal intelligence research, while the latter explores the SNR-t bias of diffusion probabilistic models.
The lukiIabs/trading-agents repository is gaining traction with a growth score of 53.94 and 233 stars, as it provides a comprehensive platform for trading agents using multi-agent finance, stocks, crypto, fintech, quantitative algo trading, and sentiment analysis. Its popularity stems from its unique approach to combining OpenAI's LLM capabilities with Node.js and JavaScript, making it an attractive solution for researchers and developers alike.
In contrast, fkyah3/opencode-yg is a research fork of opencode that demonstrates Language Anchoring, enabling LLMs to think consistently in a specific language. With 100 commits over the past 30 days and a growth score of 17.92, this project is gaining attention for its potential to improve the compliance and consistency of AI systems, particularly in languages like Chinese.
matrix-agent/awesome-agentic-world-modeling boasts an impressive 194 stars and a growth score of 8.78, despite having only 12 commits over the past month. This curated list provides a comprehensive overview of agentic world modeling, covering foundations, capabilities, laws, and beyond, making it a valuable resource for researchers in this field.
AutoMedBench/AutoMedBench is another notable project, with a growth score of 8.44 and 25 stars. As a medical auto-research benchmark for autonomous AI agents, it offers a standardized framework for evaluating the performance of AI systems in medical applications, which explains its growing popularity among researchers in this domain.
thunlp/OPD has garnered significant attention with 290 stars and a growth score of 7.43, thanks to its official repository for the paper "Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe." This project offers valuable insights into the distillation process of LLMs, making it an essential resource for researchers in this area.
Other notable projects include XIAO4579/PRISM, which explores pre-alignment via black-box on-policy distillation for multimodal RL, and gameworld-project/gameworld, a standardized evaluation framework for multimodal game agents. Both projects have demonstrated moderate growth, with PRISM scoring 5.84 and gameworld-project/gameworld scoring 5.12.
Lastly, Hedlen/Awesome-Multimodal-Intelligence and AMAP-ML/DCW are two more projects worth mentioning, despite their relatively lower growth scores of 3.68 and 3.54, respectively. The former provides a curated collection of papers, code, and datasets for multimodal intelligence research, while the latter explores the SNR-t bias of diffusion probabilistic models.