Today's AI Research: Fastest-Growing Projects — May 08, 2026
Today's AI Research, we're seeing a surge of interest in multimodal intelligence, with several repositories focused on language models, agentic world modeling, and multimodal game agents. Meanwhile, researchers are also exploring new approaches to trading, sentiment analysis, and autonomous decision-making. As the field continues to evolve, it's clear that these emerging areas will play a crucial role in shaping the future of AI.
One standout repository is lukiIabs/trading-agents (Growth Score: 70.14, Stars: 244), which provides an OpenAI-powered trading platform for stocks, crypto, and fintech applications. With its strong growth score, it's clear that researchers are eager to explore the intersection of AI and finance.
Another notable repository is fkyah3/opencode-yg (Growth Score: 20.12, Stars: 35), a research fork demonstrating Language Anchoring for consistent language thinking in LLMs. With an impressive 100 commits over the past month, it's clear that this project is actively being developed and refined.
matrix-agent/awesome-agentic-world-modeling (Growth Score: 9.86, Stars: 192) offers a comprehensive overview of agentic world modeling, covering foundations, capabilities, laws, and beyond. As researchers continue to explore the possibilities of agent-based modeling, this repository serves as a valuable resource for staying up-to-date on the latest developments.
AutoMedBench/AutoMedBench (Growth Score: 9.47, Stars: 24) provides a benchmark for autonomous AI agents in medical research, allowing developers to test and evaluate their models in a standardized environment. With its strong growth score and frequent commits, it's clear that this repository is gaining traction among researchers.
thunlp/OPD (Growth Score: 7.31, Stars: 257) offers an official implementation of on-policy distillation for large language models, providing insights into the phenomenology, mechanism, and recipe for successful knowledge transfer. As one of the most-starred repositories on this list, it's clear that researchers are eager to learn from this work.
XIAO4579/PRISM (Growth Score: 6.32, Stars: 60) explores pre-alignment via black-box on-policy distillation for multimodal RL, pushing the boundaries of what is possible in reinforcement learning. With its strong growth score and active development, this repository is definitely worth keeping an eye on.
gameworld-project/gameworld (Growth Score: 5.52, Stars: 170) provides a standardized environment for evaluating multimodal game agents, allowing researchers to compare and contrast different approaches to game-playing AI. As interest in game-based AI research continues to grow, this repository is likely to remain an important resource.
Hedlen/Awesome-Multimodal-Intelligence (Growth Score: 4.17, Stars: 37) offers a curated collection of papers, code, and datasets related to multimodal intelligence, covering VLMs, VLAs, world models, and embodied AI. While its growth score may be lower than some other repositories on this list, it's clear that this collection is providing valuable insights for researchers.
AMAP-ML/DCW (Growth Score: 3.86, Stars: 114) explores the SNR-t bias of diffusion probabilistic models, shedding light on the underlying mechanisms of these powerful tools. With its strong growth score and active development, it's clear that this repository is making important contributions to the field.
Yovecent/UDM-GRPO (Growth Score: 3.48, Stars: 21) provides a stable and efficient approach to group relative policy optimization for uniform discrete diffusion models, offering insights into the possibilities of reinforcement learning in complex environments. With its strong growth score and active development, this repository is definitely worth keeping an eye on.
One standout repository is lukiIabs/trading-agents (Growth Score: 70.14, Stars: 244), which provides an OpenAI-powered trading platform for stocks, crypto, and fintech applications. With its strong growth score, it's clear that researchers are eager to explore the intersection of AI and finance.
Another notable repository is fkyah3/opencode-yg (Growth Score: 20.12, Stars: 35), a research fork demonstrating Language Anchoring for consistent language thinking in LLMs. With an impressive 100 commits over the past month, it's clear that this project is actively being developed and refined.
matrix-agent/awesome-agentic-world-modeling (Growth Score: 9.86, Stars: 192) offers a comprehensive overview of agentic world modeling, covering foundations, capabilities, laws, and beyond. As researchers continue to explore the possibilities of agent-based modeling, this repository serves as a valuable resource for staying up-to-date on the latest developments.
AutoMedBench/AutoMedBench (Growth Score: 9.47, Stars: 24) provides a benchmark for autonomous AI agents in medical research, allowing developers to test and evaluate their models in a standardized environment. With its strong growth score and frequent commits, it's clear that this repository is gaining traction among researchers.
thunlp/OPD (Growth Score: 7.31, Stars: 257) offers an official implementation of on-policy distillation for large language models, providing insights into the phenomenology, mechanism, and recipe for successful knowledge transfer. As one of the most-starred repositories on this list, it's clear that researchers are eager to learn from this work.
XIAO4579/PRISM (Growth Score: 6.32, Stars: 60) explores pre-alignment via black-box on-policy distillation for multimodal RL, pushing the boundaries of what is possible in reinforcement learning. With its strong growth score and active development, this repository is definitely worth keeping an eye on.
gameworld-project/gameworld (Growth Score: 5.52, Stars: 170) provides a standardized environment for evaluating multimodal game agents, allowing researchers to compare and contrast different approaches to game-playing AI. As interest in game-based AI research continues to grow, this repository is likely to remain an important resource.
Hedlen/Awesome-Multimodal-Intelligence (Growth Score: 4.17, Stars: 37) offers a curated collection of papers, code, and datasets related to multimodal intelligence, covering VLMs, VLAs, world models, and embodied AI. While its growth score may be lower than some other repositories on this list, it's clear that this collection is providing valuable insights for researchers.
AMAP-ML/DCW (Growth Score: 3.86, Stars: 114) explores the SNR-t bias of diffusion probabilistic models, shedding light on the underlying mechanisms of these powerful tools. With its strong growth score and active development, it's clear that this repository is making important contributions to the field.
Yovecent/UDM-GRPO (Growth Score: 3.48, Stars: 21) provides a stable and efficient approach to group relative policy optimization for uniform discrete diffusion models, offering insights into the possibilities of reinforcement learning in complex environments. With its strong growth score and active development, this repository is definitely worth keeping an eye on.