Mischief 1 API and Pricing, Explained Through a Real Pipeline
Mischief 1 is TBC's image to video model, and at half a cent per second it is the cheapest video model in our catalog by a wide margin. It only animates a still you give it, renders 3 or 5 seconds at 16:9, and comes back silent. Here is the rate card, what our first render measured, what the endpoint refuses, and how to reach it as a versioned API endpoint.
Mischief 1 is an image to video model from TBC, The Biological Computing Co. It reached OpenRouter on 29 October 2026 and went onto Treza's canvas and into chat the next morning. The reason to pay attention is the price: half a cent per second of video at 480p, one cent at 720p. Nothing else in our video catalog comes close.
The reason to read past the price is the shape of the model. Mischief 1 does not render from a prompt alone. It animates a still you hand it, for 3 or 5 seconds, in 16:9, with no sound. This post covers the rate card, what our first render measured, what the endpoint refuses, and how to call Mischief 1 as a published API endpoint. Every parameter and price below is read off Treza's live video model catalog as of 30 October 2026, and the behavior was checked with real renders before the model went on the canvas.
What you get before you talk about price
| Capability | Mischief 1 |
|---|---|
| Input | A first frame image plus a prompt. A prompt alone is refused |
| Clip length | 3 or 5 seconds |
| Aspect ratio | 16:9 only |
| Resolution | 480p or 720p, 480p when none is set |
| Audio | None |
| Keyframes | First frame only |
| Reference images | No |
Read that table as a list of decisions the model has already made for you: widescreen, short, silent, and always starting from a picture. If the shot you need is a vertical Reel, a 10 second establishing shot, or a clip with dialogue, Mischief 1 is the wrong node. If it is a still you already have that needs to move for a few seconds, it is the cheapest way to get there. Capability detail lives on the Mischief 1 model page.
How Mischief 1 bills
Per second of output, at a rate set by the resolution:
| Resolution | Per second | 3 second clip | 5 second clip |
|---|---|---|---|
| 480p | $0.005 | $0.015 | $0.025 |
| 720p | $0.01 | $0.03 | $0.05 |
Those are the listed rates on the endpoint we run. The most expensive clip the model can render, 5 seconds at 720p, is five cents of provider cost. Twenty of them is one dollar.
For scale, the next cheapest video models in the catalog are HeyGen Video and Grok Imagine 1.5 Lite at 2 cents per second at 480p, four times the Mischief 1 rate, and Veo 3.1 Lite at 3 cents per second for a silent 720p clip. The per-second figures across the catalog are collected in the AI video cost breakdown.
What one render actually measured
Before a model goes in the picker we run it through the node's own request shape and read the bill. For Mischief 1 the first frame was a product photo of a folded navy shirt on a wooden table, and the prompt asked for a barista pouring steamed milk into a latte, a slow push-in, shallow depth of field and natural ambient sound.
What came back, in 19 seconds:
- A 3 second clip at 832 by 480 pixels, 16 frames per second, H.264.
- The shirt and the table exactly where the photo put them, with a steel pitcher pouring into a latte cup behind them. The model kept the still's composition and added the prompt's action around it.
- No audio track at all, despite the prompt asking for ambient sound. Mischief 1 is picture only.
- A charge of exactly $0.015, the listed rate times three seconds.
A second render with no Resolution set came back at the same 832 by 480 and the same $0.015 in 15 seconds, which is how we know 480p is the default. We have not rendered 720p or a 5 second clip, so we cannot say what frame size 720p returns. The 16 frames per second is worth knowing: it is a lower frame rate than most video models return, so fast motion will look choppier than the same move on a 24 frame model.
The input it refuses
The same probe sent the model a prompt with no image. The endpoint answered with a 400 error before anything rendered, saying it requires exactly one frame image of type first frame. A refused submit cannot be charged, so finding that out cost nothing.
On Treza that refusal happens one step earlier. The Video Generation node carries a limitation note for Mischief 1 that says to connect an image, and chat refuses the request before any spend when you ask for Mischief 1 without an opening frame. If you want text to video at the budget end, pick a model that renders from a prompt: Veo 3.1 Lite is the usual budget pick, and Wan 3.0 when the clip needs to run longer.
The other way to use Mischief 1 from text is to make the frame first. Put an Image Generation node ahead of it, wire its output into the video node's first frame input, and describe the motion in the video prompt. An image on a flash-tier model costs a few cents, so the two-node shot still lands under a dime. The AI image generator page has that image side set up, and the first frame post walks through the technique on the video side.
Where the still comes from
Because every Mischief 1 clip opens on your image, the quality of the image is the quality of the shot. Two sources work well. A product photo, where the photo stays put and the prompt adds a camera move or a small event around it, which is what the image to video generator is built for. Or a generated frame, where an image model decides the look and Mischief 1 only has to animate it, the easier half of the job. Either way, the prompt should describe what changes, not what is already in the picture. The model has the picture.
What you actually pay on Treza
Mischief 1 through a Treza pipeline is metered from a prepaid credit balance. One credit equals $0.01, credit packs start at $5, credits never expire, and a failed generation charges nothing.
Before a run starts, the platform estimates the media cost of the graph and refuses to start one your balance cannot cover. For Mischief 1 the estimator holds the 720p rate, one cent per second, whatever resolution the node is set to. A 3 second clip is held at 3 cents and a 480p render settles at 1.5 cents, with the difference returned to the balance when the run settles. Reserves are holds, not prices.
Calling Mischief 1 as an API without building the integration
Direct API access to a video model means credentials, a submit-and-poll job pattern, retries, a 400 to handle when the image is missing, and somewhere to store the output.
On a Treza AI video pipeline, Mischief 1 is a node on a canvas. Wire an image into its first frame input, set the duration and resolution, put it next to an Image Generation step or a Text to Speech step, and publish the whole chain as one versioned endpoint. Your application calls a typed /invoke route with JSON in and JSON out, or points an existing OpenAI SDK at the OpenAI-compatible route.
When a better image to video model lands, you change the dropdown and republish, and the endpoint your product calls does not change. Our hosted MCP server exposes the same pipelines as tools, so an agent in Claude or ChatGPT can animate a still and return the link.
Where Mischief 1 earns its place in a graph
Three jobs where the price matters more than the limits.
Draft passes. Animate a dozen candidate stills at 480p for 15 cents total, pick the one that moves well, and send that still to a higher-resolution model with sound for the final.
Product loops. A row of catalog photos, each turned into a 3 second 16:9 move, is the kind of volume job where the price per second decides whether it happens at all.
Cutaways in a longer piece. A narrated video built from generated stills can give each still a few seconds of motion for a cent or two per shot, then let the Sequence node stitch up to 12 shots over the voiceover, since Mischief 1 brings no audio of its own.
The fastest way to learn what it costs for your own work is to run your real still at your real duration and read the charge. Open the AI video generator, pick Mischief 1, drop in an image, and generate once.
Frequently Asked Questions
How much does Mischief 1 cost per second?
$0.005 per second at 480p and $0.01 per second at 720p, the listed rates on the endpoint we run as of 30 October 2026. A 3 second 480p clip is 1.5 cents of provider cost and a 5 second 720p clip is 5 cents. Our 3 second test render billed exactly the listed rate, and on Treza a failed generation charges nothing.
Can Mischief 1 generate video from text alone?
No. The endpoint requires exactly one first frame image and refuses a prompt without one before anything renders. On Treza, chat refuses the request before any spend. To start from text, generate the frame with an Image Generation node first, or pick a text to video model such as Veo 3.1 Lite or Wan 3.0.
What resolution does Mischief 1 render?
480p or 720p. With no resolution set it renders 480p, which came back as 832 by 480 pixels at 16 frames per second in our test. We have not rendered 720p yet.
How long can a Mischief 1 clip be?
3 or 5 seconds, in 16:9 only. For longer pieces, stitch clips with the Sequence node, which handles up to 12 shots, or assemble them in the timeline editor.
Does Mischief 1 generate audio?
No. Our render asked for ambient sound in the prompt and came back with no audio track. Plan sound in the graph with a Text to Speech node or an Audio Generation node, mixed over the picture in the Sequence node.
Can I use Mischief 1 in Treza chat?
Yes. Ask for Mischief 1 by name and give it an image to animate. Without an image, chat tells you the model needs an opening frame rather than starting a run.
Can I call Mischief 1 through an API without a TBC account?
Yes. On a Treza pipeline, Mischief 1 is a node you configure once and publish as a versioned HTTP endpoint, with a typed /invoke route and an OpenAI-compatible route. Generations are metered from prepaid credits that start at $5 and never expire.
