Perceptron Mk1.5
Perceptron Mk1.5 is an embodied reasoning model developed by Perceptron designed for physical agents. It processes text, image, video, and audio inputs to generate text responses and structured spatial annotations.
Perceptron Mk1.5 specs and pricing
- Provider: Perceptron
- Released: September 25, 2026
- Context window: 37K tokens
- Max output: 8K tokens
- Input: text, images, video, audio
- Output: text
- Price: $0.15 input / $1.50 output per 1M tokens
- Reasoning: yes
- Tool calling: yes
What Perceptron Mk1.5 is good at
- Analyzing video and audio data for physical agent navigation
- Extracting structured spatial data like points and polygons from images
- Reasoning-heavy tasks that require multi-modal input processing
To-dos to try with Perceptron Mk1.5
- Analyze this video clip and provide a list of bounding box coordinates for every moving vehicle.
- Listen to this audio recording and describe the physical environment based on the ambient sounds.
- Examine this image and generate a polygon annotation for the damaged area on the machine component.
FAQ
- What types of data can Perceptron Mk1.5 process? The model accepts text, image, video, and audio inputs.
- What kind of output does this model provide? It generates text responses and can include structured annotations such as points, boxes, polygons, and tracks.
- How much does Perceptron Mk1.5 cost? Perceptron Mk1.5 costs $0.15 per million input tokens and $1.50 per million output tokens through OpenRouter. In todo.is a typical agent step costs about 0.63 credits ($1 = 100 credits).
- What is the context window of Perceptron Mk1.5? Perceptron Mk1.5 can read up to 37K tokens at once and write up to 8K tokens in one reply.
- Can I use Perceptron Mk1.5 without coding? Yes. In todo.is you pick Perceptron Mk1.5 as your agent's model and write to-dos in plain words, on the web, the Android app, WhatsApp or Telegram. No API key needed: credits are included, or you can bring your own key.
JavaScript is required to use the todo.is app.