Speechmatics - API for Voice AI Agents
Mini toolsPerceptron Mk1
Perceptron Mk1, or Mark One, is a video and vision-language model from Perceptron Inc. built for physical-world AI, video understanding, and embodied reasoning. Perceptron describes it as purpose-built for video understanding and embodied reasoning, while OpenRouter lists it as accepting image and video inputs paired with natural-language queries and returning detailed visual understanding responses in either structured or natural language. It is positioned for tasks such as temporal video analysis, spatial reasoning, event understanding, and embodied perception workflows, with OpenRouter listing a 32,768-token context window and text output.
New
Multimodal
Gen 3
Overview
Perceptron Mk1 is Perceptron’s closed-source vision-language model for video understanding, image reasoning, and embodied reasoning tasks.
About Perceptron AI
Perceptron AI is an American physical AI research lab building foundation models for the physical world. Its Mk1 model delivers promptable real-time visual intelligence and spatial reasoning across manufacturing, construction, robotics, sports, and media — matching frontier lab performance at a fraction of the cost. Available via API and OpenRouter. Also open-sourcing the Isaac family of vision-language models (2B/1B). Founded November 2024 by Armen Aghajanyan and Akshat Shrivastava.
View Company ProfileTools using Perceptron Mk1
No tools found for this model yet.
Last updated: May 22, 2026
