Perceptron: Perceptron Mk1

Perceptron Mk1 is a paid model from Perceptron that accepts text, image, and video inputs, giving it broader multimodal reach than text-only alternatives. Its context window is 32,768 tokens with a maximum completion length of 8,192 tokens. The model supports reasoning and does not support tool use or structured output, which limits its suitability for agentic pipelines or applications that require reliable JSON schemas. At $0.15 per million input tokens and $1.50 per million output tokens, Perceptron Mk1 sits at a low input cost, though the output price warrants attention for high-volume generation workloads. The more significant caveat is that it carries no independent benchmark coverage, so there is no third-party data to validate its reasoning quality or compare it against peers. Teams willing to run their own evaluations, particularly those needing video understanding at a modest input rate, have reason to shortlist it; others should treat its capabilities as unproven until coverage improves.