A tiny software layer from lab-grown neurons promises faster, cheaper AI video
TBC plans a neuron-inspired video adapter, but the commercial model and performance claims remain unverified.
TBC, formerly Biological Black Box, says measurements from lab-grown cortical neurons on 4,096-electrode chips are turned into small software adapters for diffusion video models. The neurons never leave the lab, and the adapter adds less than 0.1 percent to model size while running on standard GPUs. On the open Oasis Minecraft world model, a roughly 156,000-parameter adapter scored about 19 percent better on differential entropy than the base model across ten test videos. TBC plans to offer a commercial model on AWS Trainium through SageMaker and AWS Marketplace, but only early access is available and the base model and evaluation details are undisclosed.
- TBC raised $25 million in February 2026.
- The Oasis adapter has about 156,000 parameters and does not change the base architecture.
- A larger adapter helped raise frame rate from about two to nearly ten fps.
- Bluesky Compute reported about 4.4 times lower inference cost in one test.
- The commercial model, hardware, and quality metrics have not been disclosed.
- Vendors
- Amazon
- Organizations
- TBCBluesky ComputeAmazon
- AI models
- Oasis
From the original article
5 min read at the-decoder.com
The Biological Computing Co. wants to team up with AWS to sell a text-to-video model that's supposed to run five times faster and 80 percent cheaper thanks to a software layer derived from real nerve cells. The tiny layer adds less than 0.
By Maximilian Schreiner · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.