TBC Mischief: What The Biological Computing Co.’s Video Model Is
Last updated: 2026-11-04
ClipWhim is an independent product and is not affiliated with, endorsed by or sponsored by The Biological Computing Co. (TBC). “Mischief 1” is used only to describe the third-party model ClipWhim runs, accessed via API; the name belongs to its owner.
TBC Mischief – officially "TBC Mischief 1" – is an AI video model from The Biological Computing Co. (TBC), a San Francisco start-up. It turns a still image into a short video and was released on October 30, 2026. What makes it unusual is TBC's claim that it is "neuron-derived": shaped by what the company learned from studying living brain cells in its lab. This page explains what that means in plain language, what is documented, and what has not been independently checked. You can try the model yourself in our Mischief 1 generator.
Who is The Biological Computing Co.?
According to The Next Web, TBC is a San Francisco company that grows networks of neurons in the lab, studies how they respond to signals, and turns what it learns into software that runs on ordinary AI chips (The Next Web). Its co-founder and chief operating officer, Jon Pomeraniec, told TNW: "We're building a platform that can change the economics of AI at scale."
TNW also reports that the launch followed a September deal with Amazon Web Services to sell TBC's models to businesses. OrigoTech, citing The Decoder, adds that as of September 22 the AWS collaboration was an early-access sign-up for a text-to-video model whose base model was not named – separate from the Mischief 1 image-to-video API (OrigoTech).
We could not confirm an official TBC website or app address from the sources above, so we do not link one here. OrigoTech mentions that the model is also available in "TBC's creation app".
What TBC Mischief 1 actually is
| Fact | Detail | Source |
|---|---|---|
| Type | Image-to-video (first frame + prompt) | OpenRouter |
| Released | October 30, 2026 | OpenRouter |
| Base model | Alibaba's open Wan 2.2 | TNW |
| Output | 480p or 720p, 16:9, 16 fps, 3 or 5 s, no audio | fal |
| API price | $0.005/s (480p), $0.01/s (720p) | fal |
| Where to use it | fal, OpenRouter, TBC creation app | OrigoTech |
OpenRouter describes it as "the world's first neuron-derived video model" and suggests it for ideation, previsualization, moodboarding and variant testing (OpenRouter). "World's first" is TBC's own marketing description.
The neuron-derived claim, explained plainly
It is easy to misread "neuron-derived" as "runs on brain cells". It does not. Every source we checked agrees on this point: the neurons stay in TBC's laboratory, and the product is ordinary software that runs on standard GPUs (TNW, OrigoTech).
Here is the chain as TBC describes it, according to OrigoTech's summary of reporting by The Decoder and Unite.AI:
- TBC grows cortical neurons on chips with thousands of electrodes and stimulates them.
- Researchers measure how activity spreads and then fades – it stays mostly local, then dies down, like a damper.
- That rule is encoded into a small software component (TBC calls an earlier version a "Neural Dynamics Adapter"). The cells are not used at runtime.
- For the model, Unite.AI (quoting TBC) describes a post-training layer around Wan 2.2's diffusion transformer blocks that adds under 0.1% to the model's parameters.
The speed and cost claims – and what is verified
TBC says the model runs about five times faster than Wan 2.2 at the same quality, and press material reported an inference cost about 80% lower. OrigoTech notes these are one effect expressed two ways (1 − 1/5 = 80%), not two separate gains.
None of this has been independently verified. The Next Web states that TBC has not published independent tests to back up the speed claim. OrigoTech adds that the base model variant, hardware, resolution, clip length and quality measure behind the figures have not been disclosed, and that no controlled comparison against purely digital speed-ups (like distillation or caching) on the same base model has been published. Whether the neuron research is what actually explains the speed-up cannot be checked from outside.
What can be checked is the price. fal lists $0.005 per second at 480p and $0.01 at 720p, and Treza Labs reports its 3-second test render billed exactly $0.015 (Treza Labs). OpenRouter currently ranks it first of 27 video models on both cost and speed on its own listing (OpenRouter) – that is a platform measurement of latency, not a quality benchmark.
We have not run our own controlled benchmarks either, so we do not add any speed or quality numbers of our own.
TBC Mischief vs the "Mischief" prank app
There is an unrelated consumer app called Mischief for making AI prank videos on iOS and Android. It has nothing to do with TBC or the model. If you searched for "mischief ai video" and wanted prank videos, that app is what you are looking for. If you want to animate your own images with TBC's model, you are in the right place.
How to try TBC Mischief 1
Developers can call it on fal or OpenRouter with an API key. If you prefer not to set up an API account, our online generator runs the model for you with credits. Start with the image-to-video guide and the prompt examples, check pricing, or compare alternatives.
Frequently asked questions
- What does TBC stand for?
- TBC is The Biological Computing Co., a San Francisco start-up that released the Mischief 1 image-to-video model on October 30, 2026.
- Does Mischief 1 run on living neurons?
- No. According to TBC as reported by The Next Web and OrigoTech, the neurons stay in the lab. The model is a software layer on top of Wan 2.2 that runs on standard GPUs.
- Is Mischief 1 really five times faster than Wan 2.2?
- That is TBC’s claim. No independent benchmark had been published at the time of writing, and the test conditions have not been disclosed.
- Is TBC Mischief related to the Mischief prank app?
- No. The Mischief AI prank video app on iOS and Android is a separate product and is unrelated to The Biological Computing Co.
- Is ClipWhim the official TBC site?
- No. ClipWhim is an independent website that provides access to the model through a third-party API. It is not affiliated with or endorsed by TBC.
Try the AI video generator
Upload an image, write a motion prompt and get a 3–5 second widescreen clip. Failed generations are refunded automatically.
Generate a videoMore guides
- Image-to-video tutorialStep-by-step guide and tips for choosing first frames.
- Pricing & cost calculatorAPI cost per second, our credits, and a batch calculator.
- Comparison with Wan 2.2Specs, prices and claims side by side.
- Alternative video modelsWan, Kling, Seedance, Veo and when to switch.
- Prompt library19 example prompts by use case, plus do’s and don’ts.