生物计算公司发布 API 借鉴活体神经元优化视频模型
BioComputing Co. 推出名为“Mischief 1”的 API，将实验室中活体人脑神经元在电极上的信号处理机制应用于开源视频生成模型 Wan 2.2。该方案通过增加不到 0.1% 的参数层，声称在保持原有质量的同时实现 5 倍推理加速，并作为最便宜的图生视频模型上线主流平台。
报道正文
Rohan Paul (@rohanpaul_ai) · 2026-10-30 14
Truly Wild stuff: a lab grew living brain cells, studied how they think, and shipped what it learned as an API you can call today.
TBC mischief 1 is that API: findings from neurons on electrodes, applied to an existing image-to-video model (Wan 2.2) .
mischief 1 applies a layer of this kind to Wan 2.2, adding less than 0.1% to its parameters, and the company claims it matches Wan 2.2’s quality at 5x the speed.
@BioComputingCo grows living human brain cells on chips with about 4,000 electrodes, where the neurons form their own networks, receive images as electrical stimulation, and their responses are decoded to improve AI models.
The path from dish to endpoint:
Grow neurons on electrodes Measure how they process information Take an existing open-weight video model and change it based on those observations Ship the model, with no cells inside
↳ 引用 @AlexKsendzovsky (Alex Ksendzovsky): 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.
