Updated

Perceptron Mk1.5

Perceptron released Mk1.5, a model for controlling embodied agents including drones, robotic dogs, smart glasses, and phones. It adds native audio input, video object tracking with timestamped geometries, and egocentric video understanding, with a 32K token context window.

By No Code MBA · Facts checked

Provider
Perceptron
Announcement date
Status at last check
Generally available
Model family
Perceptron Mk1
Preceding model
Perceptron Mk1
Capabilities
Text, Image, Video, Audio

What changed

Mk1.5 adds native audio input, object tracking across video returned as timestamped geometries, egocentric video understanding, tool use with OpenAI-format function definitions, and sub-agent calls. It accepts text, images, video and audio with a 32K token context, and outputs text, points, boxes, polygons, clips and tracks. Perceptron reports up to 4.7x faster end-to-end chat than Mk1 on one H100 and 80.40% on EgoSchema.

Access and availability

Available through the Perceptron Platform and updated SDK using the model ID perceptron-mk1.5; on-premises deployment is offered through sales.

Official access documentation

Pricing

$0.15 per million input tokens and $1.50 per million output tokens

Official pricing

What you can build

Try it through the Perceptron Platform API at $0.15/M input tokens for tasks that require real-time video tracking or embodied agent control without platform-specific retraining.

Limitations

Perceptron says it is not at the frontier of audio-visual understanding and trails Gemini 3.1 Pro on some video benchmarks such as Perception Test. It relies on tool calls rather than memorized world knowledge, and the 32K context is small.

Sources and release history