MODELS
Perceptron: Perceptron Mk1
by Perceptron AI
Overview
Perceptron Mk1 is a Perceptron AI vision-language model for image and video understanding with reasoning support.
Details
Perceptron Mk1, also described as Mark One, is Perceptron's highest-quality vision-language model for video and embodied reasoning. The documentation describes it as a vision-language model for images and video with model ID `perceptron-mk1`, 32K context, reasoning support, closed-source status, and pricing of $0.15 per million input tokens and $1.50 per million output tokens. The official Python SDK repository says it supports Perceptron Mk1 for grounded perception tasks including detection, localization, OCR, and visual Q&A.
When to Use
Use Perceptron Mk1 when an application needs image or video inputs paired with natural language queries for detailed visual understanding. Use it for grounded perception tasks referenced by the official SDK, such as detection, localization, OCR, and visual Q&A. Consider it when evaluating a closed-source commercial vision-language model with documented reasoning support and a 32K context window.
Getting Started
- Review the Perceptron Mk1 quickstart in the Perceptron documentation.
- Use the model ID `perceptron-mk1` when following the documented API flow.
- Install or inspect the official Python SDK at `perceptron-ai-inc/perceptron` if you want SDK-based integration.
- Check the documented pricing of $0.15/M input and $1.50/M output before running workloads.
Key Features
- •Vision-language model for image and video inputs
- •Natural-language querying over visual inputs
- •Reasoning support
- •32K context window
- •Supported by Perceptron AI's official Python SDK
- •Grounded perception task support including detection, localization, OCR, and visual Q&A
Capabilities
- •image-understanding
- •video-understanding
- •visual-question-answering
- •ocr
- •object-detection
- •localization
- •reasoning
Last updated May 31, 2026