Perceptron AI
1 model built · 1 served
Perceptron AI is a physical-AI lab founded in 2024 by ex-Meta FAIR researchers, based in the United States. Rather than chasing general-purpose chat models, they're focused on embodied AI and video understanding — the kind of work that matters for robotics and systems that need to reason about the physical world from visual input. Their only released model so far, Perceptron Mk1, reflects that focus: it takes text, image, and video input and outputs text, which fits a lab building toward physical-world understanding rather than another chatbot. One model in, this is a lab to watch rather than one with a broad track record yet.
Lineup
With a single model in the lineup, the choice is simple: use Perceptron Mk1. It's a text+image+video-to-text model priced at $0.15 per 1M input tokens and $1.50 per 1M output tokens (OpenKey price: $0.1545 in / $1.545 out, at provider price x 1.03), with a 32,768-token context window. That price point puts it in the cheap-for-experimentation range, especially for anything involving video input, so it's a reasonable pick if you're testing embodied or video-understanding workloads and don't need a huge context window.
- Perceptron Mk1May 12, 2026 33K $0.154 $1.54 ctx 33K · in $0.154 · out $1.54
Prices per 1M tokens, flat 3% fee included.
Models Perceptron AI serves
Beyond its own models, Perceptron AI hosts 1 model in the OpenKey catalog. Prices are Perceptron AI’s own list price per 1M input tokens, before our fee.
| Model | Built by | Context | Input /M |
|---|---|---|---|
| Perceptron Mk1 | Perceptron AI | 33K | $0.150 |
From the OpenRouter catalog, Jul 28, 2026. All providers
Privacy policy ↗Terms of service ↗
Questions
- What does Perceptron Mk1 cost on OpenKey?
- Perceptron Mk1 is $0.15 per 1M input tokens and $1.50 per 1M output tokens at provider list price. On OpenKey that's $0.1545 input / $1.545 output per 1M tokens — provider price plus the flat 3% fee, no separate markup tiers.
- Which Perceptron model should I use?
- There's only one: Perceptron Mk1. It handles text, image, and video input and returns text, with a 32,768-token context window. Since it's the lab's sole released model, it's the model by default rather than a choice among options.
- Can I access Perceptron models through a single API key?
- Yes. OpenKey gives you one key across all 329 models from 52 labs, including Perceptron Mk1, so you don't need a separate account with Perceptron AI to try it alongside models from other labs.