Today's AI Frameworks & SDKs: Fastest-Growing Projects — October 03, 2026
Today's AI Frameworks & SDKs category showcases a mix of projects that cater to various aspects of decision-making and machine learning processes, with a notable surge in repositories related to System One models and interactive 3D frameworks. The leading project, "von," has seen impressive growth, likely due to its innovative approach to non-autoregressive decision modeling.
wfzyx/von
The von framework offers an open-source implementation of the TypeSafe Jev model, providing a local drop-in alternative with sub-15ms performance. Its high growth score and star count indicate that developers are drawn to its promise of efficient, rapid decision-making capabilities without the need for cloud-based services.
AbdelStark/awesome-typesafe-jev
This repository serves as a comprehensive guide to TypeSafe's System One model, including SDKs, live demos, agent tools, and independent evaluations. Its robust collection of resources and high level of community engagement suggest that it is becoming an essential reference for those working with the System One decision model.
AnotiaWang/awesome-decision-models
Curating a wide range of decision models, this repository includes hosted APIs, open-weight models, runtimes, SDKs, applications, benchmarks, and research papers. The significant growth in stars and commits highlights its importance as a go-to resource for developers seeking comprehensive information on various decision-making frameworks.
PSRben/VisionHOPE
VisionHOPE provides an official PyTorch implementation of Visual Backbones as Self-Modifying Learning Systems. Its growing popularity may be due to the increasing interest in self-modifying learning systems within the machine learning community, which seeks more adaptive and efficient models for visual tasks.
deepseek-ai/DeepEP-Ascend
This project offers a high-performance communication library designed specifically for training and inference on Huawei Ascend NPUs. The modest growth but steady engagement suggest that it is gaining traction among developers working with specific hardware solutions for machine learning applications.
blendi-remade/dioramas
Dioramas is an open-source framework that generates cinematic, interactive 3D websites using AI-generated assets. Its growing star count reflects the increasing interest in leveraging AI for creative and immersive web experiences, making it a notable tool for developers looking to enhance their projects with dynamic visuals.
CaptureGrubEnchant/SolidWorks
SolidWorks MCP Server integrates an AI assistant into SolidWorks CAD software, enabling tasks like sketching, extruding, filleting, exporting STEP/STL files, and generating macros. The substantial growth in stars indicates that its utility for enhancing the capabilities of CAD software is resonating with a wide audience.
maanHimself/OpenDLSS-NR
This project offers a Vulkan reimplementation of NVIDIA's DLSS 5 Neural Rendering network, ensuring bit-exact performance against the original implementation. Its high star count and steady growth suggest that developers are keen on exploring open-source alternatives for neural rendering capabilities in gaming and graphics applications.
MiniMax-AI/OpenAgentCore
OpenAgentCore provides a self-hosted implementation of the OpenAI Agents API with multiple native harnesses, enabling local execution of AI agents. The growing interest in self-hosting solutions likely drives its popularity as developers seek more control over their AI environments.
ThinkWatchProject/ThinkWatch-Lite
This project offers a local gateway for various AI clients like Claude Code and Codex, allowing users to manage API keys securely and trace requests without interfering with client software. Its significant growth in stars reflects the growing demand for secure and flexible middleware solutions that enhance privacy and control over AI interactions.
The diversity of projects this week underscores the expanding horizons of AI frameworks and SDKs across decision-making, visual learning systems, creative web development, CAD integration, neural rendering, self-hosted agents, and security management.
wfzyx/von
The von framework offers an open-source implementation of the TypeSafe Jev model, providing a local drop-in alternative with sub-15ms performance. Its high growth score and star count indicate that developers are drawn to its promise of efficient, rapid decision-making capabilities without the need for cloud-based services.
AbdelStark/awesome-typesafe-jev
This repository serves as a comprehensive guide to TypeSafe's System One model, including SDKs, live demos, agent tools, and independent evaluations. Its robust collection of resources and high level of community engagement suggest that it is becoming an essential reference for those working with the System One decision model.
AnotiaWang/awesome-decision-models
Curating a wide range of decision models, this repository includes hosted APIs, open-weight models, runtimes, SDKs, applications, benchmarks, and research papers. The significant growth in stars and commits highlights its importance as a go-to resource for developers seeking comprehensive information on various decision-making frameworks.
PSRben/VisionHOPE
VisionHOPE provides an official PyTorch implementation of Visual Backbones as Self-Modifying Learning Systems. Its growing popularity may be due to the increasing interest in self-modifying learning systems within the machine learning community, which seeks more adaptive and efficient models for visual tasks.
deepseek-ai/DeepEP-Ascend
This project offers a high-performance communication library designed specifically for training and inference on Huawei Ascend NPUs. The modest growth but steady engagement suggest that it is gaining traction among developers working with specific hardware solutions for machine learning applications.
blendi-remade/dioramas
Dioramas is an open-source framework that generates cinematic, interactive 3D websites using AI-generated assets. Its growing star count reflects the increasing interest in leveraging AI for creative and immersive web experiences, making it a notable tool for developers looking to enhance their projects with dynamic visuals.
CaptureGrubEnchant/SolidWorks
SolidWorks MCP Server integrates an AI assistant into SolidWorks CAD software, enabling tasks like sketching, extruding, filleting, exporting STEP/STL files, and generating macros. The substantial growth in stars indicates that its utility for enhancing the capabilities of CAD software is resonating with a wide audience.
maanHimself/OpenDLSS-NR
This project offers a Vulkan reimplementation of NVIDIA's DLSS 5 Neural Rendering network, ensuring bit-exact performance against the original implementation. Its high star count and steady growth suggest that developers are keen on exploring open-source alternatives for neural rendering capabilities in gaming and graphics applications.
MiniMax-AI/OpenAgentCore
OpenAgentCore provides a self-hosted implementation of the OpenAI Agents API with multiple native harnesses, enabling local execution of AI agents. The growing interest in self-hosting solutions likely drives its popularity as developers seek more control over their AI environments.
ThinkWatchProject/ThinkWatch-Lite
This project offers a local gateway for various AI clients like Claude Code and Codex, allowing users to manage API keys securely and trace requests without interfering with client software. Its significant growth in stars reflects the growing demand for secure and flexible middleware solutions that enhance privacy and control over AI interactions.
The diversity of projects this week underscores the expanding horizons of AI frameworks and SDKs across decision-making, visual learning systems, creative web development, CAD integration, neural rendering, self-hosted agents, and security management.