MODELS

Perceptron Mk1

by Perceptron AI

multimodalclosed-source

Overview

Vision-language model for image and video understanding, OCR, object detection, event clipping, and embodied reasoning via API.

Details

Perceptron Mk1 is a Perceptron AI vision-language model documented for image and video understanding, OCR, object detection, event clipping, and embodied reasoning via API. The official model spec lists model ID perceptron-mk1, text/image/video inputs, text output, 32K context, reasoning support, supported MIME types, and closed-source status. The API reference documents OpenAI-compatible multimodal chat completions at POST /v1/chat/completions, including streaming, vision_config controls, 32K context, and 8K output limits. Perceptron AI’s launch release says the model is available through the Perceptron AI API platform and OpenRouter.

When to Use

Use for multimodal applications that need text image and video inputs with text output through an OpenAI-compatible chat-completions API. Use when a workflow needs documented vision-language functions such as OCR object detection video understanding event clipping or embodied reasoning. Use when 32K context streaming responses and vision_config controls are relevant to the integration.

Getting Started

Read the Perceptron Mk1 documentation index and model spec to confirm the perceptron-mk1 model ID
supported inputs
output type
MIME types
context window
and limits.
Use the API reference for POST /v1/chat/completions to structure an OpenAI-compatible multimodal request and test streaming or vision_config options if needed.
Review the Perceptron security guidance for API-key storage and rotation
PII sanitization
TLS
firewall rules
encrypted JSON exports
and protection of annotated examples.
If desired
evaluate availability through the Perceptron AI API platform or OpenRouter as described in the launch release.

Key Features

•Vision-language model for image and video understanding
•Supports OCR
•object detection
•event clipping
•and embodied reasoning use cases
•Accepts text
•image
•and video inputs with text output
•OpenAI-compatible multimodal chat-completions endpoint
•32K context window and documented 8K output limit
•Reasoning support and streaming support documented in official sources

Capabilities

•vision-language understanding
•image understanding
•video understanding
•OCR
•object detection
•event clipping
•embodied reasoning
•multimodal chat completions
•streaming

Last updated Jun 22, 2026