The Biological Computing Co. Launches Mischief 1, a Video Model It Says Draws on Living Neurons
today, we introduce a little mischief to the world. for four years, we’ve been brewing real brain cells on electrodes to understand how they compute. now, we’ve turned those discoveries into a video model anyone can use. from 0.5¢ a second, mischief 1 is at least 5x faster than its base and the cheapest image-to-video model on @fal and @OpenRouter. soon to be on @awscloud marketplace. it’s not a delusion anymore. biocompute is real, commercial and this is just the beginning. happy halloween.
Context
TBC says lessons from lab grown neurons made a popular open video model five times faster, which, if it holds up, would be an early commercial use of biological computing in AI.
The Biological Computing Co., or TBC, is a San Francisco startup that grows living neurons on electrodes and studies how they process information. Alex Ksendzovsky is its CEO and cofounder. Wan 2.2 is a widely used open weight video model, and fal and OpenRouter are marketplaces where developers call AI models.
Deep dive
Alex Ksendzovsky, CEO of The Biological Computing Co., said on Friday that the company spent four years growing real brain cells on electrodes to learn how they compute and has now turned those discoveries into a video model anyone can use. These are the company's claims.
According to the post, Mischief 1 costs from half a cent per second, is at least five times faster than its base model, and is the cheapest image to video model on fal and OpenRouter, with an AWS Marketplace listing coming soon.
TBC's press release says the model is built on Wan 2.2 and matches its quality while generating video five times faster. It says the neurons stay in the lab, and that the discoveries become software that runs on standard AI hardware and adds less than 0.1 percent to the base model's parameter count. A five second clip costs 2.5 cents at the lowest tier.
The OpenRouter listing shows 480p and 720p output, three or five second clips, and pricing from 0.005 dollars per second at 480p to 0.01 dollars at 720p. TBC has not published independent tests showing how much of the speedup comes from neuron inspired methods rather than standard optimization. The company says it plans to apply the approach to language and physical AI next.
What to expect next
Watch for independent benchmarks comparing Mischief 1 with Wan 2.2 on speed and quality, and for technical detail on what the neuron derived methods actually change. The AWS Marketplace listing is promised soon. If TBC brings the approach to language models as planned, claims of efficiency gains without more compute will face much closer scrutiny. Developers can try the model now, so user comparisons should appear quickly.