Mininglamp Launches Cider+Mano-P, Transforming Mac into a Private AI Workstation
Recently, Mininglamp open-sourced two key local AI projects—Cider and Mano-P—that solve critical challenges in Mac inference acceleration and GUI agent operations, offering users a full local AI setup. This transforms the Mac from merely being "able to run AI" into an efficient, private, and deeply controllable AI workstation.

Cider: Unlocking M-Series Chip Potential for Faster, More Efficient Local LLM/VLM Execution
When running local large models on a Mac, many users find that despite powerful hardware, inference speed and memory use often fall short. Cider was built to address this exact problem.
This project fully taps into the INT8 TensorOps engine of M-series chips (particularly the M5), dramatically boosting local inference performance for LLMs and VLMs while slashing memory usage. It offers Mac users a tangible way to run AI models efficiently on edge devices.
Mano-P: A Pure Visual GUI Agent That Enables Full AI "Screen Viewing and Operation" Workflow
While Cider focuses on speed, Mano-P tackles usability—giving AI the ability to understand and control computers just like a human.
Mano-P is a GUI-VLA agent designed for edge devices, running locally on Mac mini and MacBook. It goes beyond browser-only agents, enabling direct interaction with desktop applications, web interfaces, professional tools, and intricate graphical workflows.
Its core capabilities include:
Complex GUI automation operations, cross-system data integration, long-task planning and execution, intelligent report generation, self-built application developmentThe system relies on pure visual GUI operations, so screenshots and task data never leave the device, ensuring complete privacy and security.
Practical Demonstration: Mano-P Learns Mahjong and Makes Autonomous Decisions
In the demo, Mano-P demonstrated strong performance in a mahjong game. Using only visual perception, it recognizes cards, analyzes the situation, and makes decisions—showing a complete "perception to action" loop.
Two Open Source Projects Join Forces to Create a Private Local AI Infrastructure
Cider (edge inference acceleration) and Mano-P (edge GUI agent) together make a complete local AI stack. Whether you want top inference speed or AI that can handle complex desktop tasks on its own, you get a more capable and private AI experience on Mac.
AIbase Comment: With the growing need for local model deployment, Mininglamp's open-sourcing of Cider and Mano-P directly addresses Mac users' pain points, offering a key reference for practical edge AI and ecosystem growth. Developers interested can check the latest updates and build their own local AI workstation.
Related article
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber
Related Special Topic Recommendations
Comments (1)
0/500
Recently, Mininglamp open-sourced two key local AI projects—Cider and Mano-P—that solve critical challenges in Mac inference acceleration and GUI agent operations, offering users a full local AI setup. This transforms the Mac from merely being "able to run AI" into an efficient, private, and deeply controllable AI workstation.

Cider: Unlocking M-Series Chip Potential for Faster, More Efficient Local LLM/VLM Execution
When running local large models on a Mac, many users find that despite powerful hardware, inference speed and memory use often fall short. Cider was built to address this exact problem.
This project fully taps into the INT8 TensorOps engine of M-series chips (particularly the M5), dramatically boosting local inference performance for LLMs and VLMs while slashing memory usage. It offers Mac users a tangible way to run AI models efficiently on edge devices.
Mano-P: A Pure Visual GUI Agent That Enables Full AI "Screen Viewing and Operation" Workflow
While Cider focuses on speed, Mano-P tackles usability—giving AI the ability to understand and control computers just like a human.
Mano-P is a GUI-VLA agent designed for edge devices, running locally on Mac mini and MacBook. It goes beyond browser-only agents, enabling direct interaction with desktop applications, web interfaces, professional tools, and intricate graphical workflows.
Its core capabilities include:
Complex GUI automation operations, cross-system data integration, long-task planning and execution, intelligent report generation, self-built application developmentThe system relies on pure visual GUI operations, so screenshots and task data never leave the device, ensuring complete privacy and security.
Practical Demonstration: Mano-P Learns Mahjong and Makes Autonomous Decisions
In the demo, Mano-P demonstrated strong performance in a mahjong game. Using only visual perception, it recognizes cards, analyzes the situation, and makes decisions—showing a complete "perception to action" loop.
Two Open Source Projects Join Forces to Create a Private Local AI Infrastructure
Cider (edge inference acceleration) and Mano-P (edge GUI agent) together make a complete local AI stack. Whether you want top inference speed or AI that can handle complex desktop tasks on its own, you get a more capable and private AI experience on Mac.
AIbase Comment: With the growing need for local model deployment, Mininglamp's open-sourcing of Cider and Mano-P directly addresses Mac users' pain points, offering a key reference for practical edge AI and ecosystem growth. Developers interested can check the latest updates and build their own local AI workstation.
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber





Home






