Today's AI Research: Fastest-Growing Projects — April 30, 2026
Today's AI Research, we're seeing a surge in innovative projects that push the boundaries of language understanding, autonomous agents, and multimodal evaluation. One notable trend is the focus on making large language models (LLMs) more consistent and controllable, as seen in several repositories that explore new approaches to LLM training and deployment.
fkyah3/opencode-yg takes the top spot with a growth score of 39.69 and 26 stars, demonstrating rapid interest in its research fork of opencode that achieves impressive results in language anchoring for LLMs, with over 95% Chinese thinking compliance verified. This project's explosive growth is likely due to its potential to make LLMs more reliable and consistent across languages.
AutoMedBench/AutoMedBench follows closely with a growth score of 18.44 and 22 stars, showcasing the MedAutoBench medical research benchmark for autonomous AI agents. As the field of healthcare continues to adopt AI-driven solutions, this repository's growth reflects the increasing demand for standardized evaluation frameworks that can assess the performance of autonomous agents in medical settings.
matrix-agent/awesome-agentic-world-modeling boasts an impressive 125 stars and a growth score of 15.42, highlighting its comprehensive survey of agentic world modeling foundations, capabilities, laws, and beyond. This repository's enduring popularity stems from its exhaustive collection of resources, making it an invaluable reference for researchers exploring the frontiers of AI agent design.
kokolerk/TCOD explores the concept of temporal curriculum in on-policy distillation for multi-turn autonomous agents, garnering a growth score of 9.75 and 21 stars. This project's moderate growth is likely driven by its novel approach to improving the efficiency and effectiveness of autonomous agents in complex environments.
gameworld-project/gameworld has attracted 163 stars and achieved a growth score of 8.23, focusing on standardized evaluation of multimodal game agents through its GameWorld framework. As AI research continues to push into gaming and interactive domains, this repository's steady growth reflects the need for robust evaluation tools that can assess agent performance across diverse scenarios.
thunlp/OPD boasts an impressive 181 stars and a growth score of 8.11, presenting the official repository for the paper "Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe". This project's continued popularity stems from its in-depth analysis of on-policy distillation techniques, offering insights into optimizing LLM training and deployment.
7WaySecurity/ai_osint has garnered 71 stars and a growth score of 8.02, providing a curated collection of AI OSINT resources for discovering exposed LLM endpoints, leaked API keys, and misconfigured vector databases. As concerns around AI security grow, this repository's steady growth reflects the increasing demand for tools that can identify vulnerabilities in AI systems.
AMAP-ML/DCW explores the SNR-t bias of diffusion probabilistic models, with a growth score of 6.12 and 111 stars. Although its growth is moderate, this project contributes to ongoing discussions around the limitations and potential biases of popular AI models, making it an important resource for researchers seeking to improve model performance.
zubair-trabzada/ai-trading-claude showcases an AI trading research engine with a growth score of 5.98 and 89 stars, providing a comprehensive analysis tool for stocks, options strategies, sector rotation, and portfolio management. This project's slow but steady growth is likely driven by its niche appeal to researchers exploring the intersection of finance and AI.
Gloriaameng/Awesome-Agent-Harness rounds out our list with a growth score of 5.59 and 92 stars, presenting a survey on LLM agent harness engineering with a taxonomy of over 110 papers and 23 systems analyzed. Although its growth is modest, this repository's enduring popularity stems from its value as a reference for researchers navigating the complex landscape of AI agent design and deployment.
fkyah3/opencode-yg takes the top spot with a growth score of 39.69 and 26 stars, demonstrating rapid interest in its research fork of opencode that achieves impressive results in language anchoring for LLMs, with over 95% Chinese thinking compliance verified. This project's explosive growth is likely due to its potential to make LLMs more reliable and consistent across languages.
AutoMedBench/AutoMedBench follows closely with a growth score of 18.44 and 22 stars, showcasing the MedAutoBench medical research benchmark for autonomous AI agents. As the field of healthcare continues to adopt AI-driven solutions, this repository's growth reflects the increasing demand for standardized evaluation frameworks that can assess the performance of autonomous agents in medical settings.
matrix-agent/awesome-agentic-world-modeling boasts an impressive 125 stars and a growth score of 15.42, highlighting its comprehensive survey of agentic world modeling foundations, capabilities, laws, and beyond. This repository's enduring popularity stems from its exhaustive collection of resources, making it an invaluable reference for researchers exploring the frontiers of AI agent design.
kokolerk/TCOD explores the concept of temporal curriculum in on-policy distillation for multi-turn autonomous agents, garnering a growth score of 9.75 and 21 stars. This project's moderate growth is likely driven by its novel approach to improving the efficiency and effectiveness of autonomous agents in complex environments.
gameworld-project/gameworld has attracted 163 stars and achieved a growth score of 8.23, focusing on standardized evaluation of multimodal game agents through its GameWorld framework. As AI research continues to push into gaming and interactive domains, this repository's steady growth reflects the need for robust evaluation tools that can assess agent performance across diverse scenarios.
thunlp/OPD boasts an impressive 181 stars and a growth score of 8.11, presenting the official repository for the paper "Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe". This project's continued popularity stems from its in-depth analysis of on-policy distillation techniques, offering insights into optimizing LLM training and deployment.
7WaySecurity/ai_osint has garnered 71 stars and a growth score of 8.02, providing a curated collection of AI OSINT resources for discovering exposed LLM endpoints, leaked API keys, and misconfigured vector databases. As concerns around AI security grow, this repository's steady growth reflects the increasing demand for tools that can identify vulnerabilities in AI systems.
AMAP-ML/DCW explores the SNR-t bias of diffusion probabilistic models, with a growth score of 6.12 and 111 stars. Although its growth is moderate, this project contributes to ongoing discussions around the limitations and potential biases of popular AI models, making it an important resource for researchers seeking to improve model performance.
zubair-trabzada/ai-trading-claude showcases an AI trading research engine with a growth score of 5.98 and 89 stars, providing a comprehensive analysis tool for stocks, options strategies, sector rotation, and portfolio management. This project's slow but steady growth is likely driven by its niche appeal to researchers exploring the intersection of finance and AI.
Gloriaameng/Awesome-Agent-Harness rounds out our list with a growth score of 5.59 and 92 stars, presenting a survey on LLM agent harness engineering with a taxonomy of over 110 papers and 23 systems analyzed. Although its growth is modest, this repository's enduring popularity stems from its value as a reference for researchers navigating the complex landscape of AI agent design and deployment.