Perceptron Mk1 Is Now Available in Puter.js

Puter.js now supports Perceptron Mk1, Perceptron AI's new vision-language model built for video understanding and embodied reasoning—priced 80–90% below comparable frontier vision-language models.

What is Perceptron Mk1?

Perceptron Mk1 is the first model from Perceptron AI, a startup founded by former Meta multimodal researchers and launched on May 12, 2026. Unlike text-first models that bolt on image inputs, Mk1 is designed from the ground up for "physical AI"—reasoning about the real world through video. Key features include:

• Native video understanding: Processes video at up to 2 frames per second within a 32K token context window, maintaining object identity and temporal continuity across frames
• Structured outputs: Returns timecodes for events, bounding boxes for objects, and point coordinates that are parseable at scale—not just free-form prose
• State-of-the-art spatial reasoning: Scores 88.5 on VSI-Bench (the highest recorded among compared models), 85.1 on EmbSpatialBench, and 72.4 on RefSpatialBench—outperforming frontier models from OpenAI, Google, and Anthropic on these benchmarks
• Frontier capabilities at a fraction of the cost: $0.15 per million input tokens, 80–90% cheaper than comparable vision-language models from major labs
• Built for real-world workloads: Video QA, event detection, open-vocabulary object detection and counting, OCR on documents and signage, hand pose estimation, and point-by-example grounding

Examples

Video question answering

puter.ai.chat(
  "Summarize what happens in this video and return timecodes for key events.",
  "https://assets.puter.site/puppy.mp4",
  { model: "perceptron/perceptron-mk1" }
);

Image understanding

puter.ai.chat(
  "What do you see in this image?",
  "https://assets.puter.site/doge.jpeg",
  { model: "perceptron/perceptron-mk1" }
);

Object detection and counting

puter.ai.chat(
  "List every object in this scene with its approximate location.",
  "https://assets.puter.site/doge.jpeg",
  { model: "perceptron/perceptron-mk1" }
);

Document OCR

puter.ai.chat(
  "Extract all text from this document and return it as plain text.",
  "https://assets.puter.site/letter.png",
  { model: "perceptron/perceptron-mk1" }
);

Get Started Now

Just add one library to your project:

// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';

Or add one script tag to your HTML:

<script src="https://js.puter.com/v2/"></script>

No API keys needed. Start building with Perceptron Mk1 immediately.

Learn more:

• Free, Unlimited Perceptron AI API
• Puter.js AI Documentation