Prodia
Lovable
Vercel
Pixlr
Kapwing
Text to video
MiniMax H3 Fast
Turn a text prompt into a video with synchronized audio.
More video.
Less spend.
Prodia is an inference API for generative video and image driven by frontier research and multi-silicon.
Test a model, then call the same model from your product at published per-job prices.
Building with a coding agent? Start with the agent guide.
- LTX 2.5 Fast$0.0465per 10-second clip
- MiniMax H3 Hyper$0.0885per 5-second clip
- MiniMax H3 Turbo$0.0902per 5-second clip
- MiniMax H3 Fast$0.2307per 5-second clip
10,000 clips, priced side by side.
The same workload at each provider's published rate. Bars show the total for 10,000 clips.
Published rates checked 10 November 2026: Prodia pricing and fal pricing.
Looking for volume pricing?
Estimate your monthly workload and ask us for a quote.
Estimate your workload
Let your agent take the first step.
Point your coding agent at the Prodia docs index. It lists every job type, parameter and reference page as Markdown. Running a job still needs your Prodia token and your go-ahead.
Models to start with.
Sample outputs from the Prodia Model Explorer. See all models
Image editing
Qwen Image Edit Plus
Change an existing image with a text instruction. From $0.0085 per edit.
Your first API request.
One endpoint for video and image jobs. The response is the generated file.
- Create an accountSign up with your email at app.prodia.com.
- Create an API tokenIn the API dashboard. Tokens require a Pro subscription. Keep yours on the server.
- Run your first generationCopy the request and save the result.
export PRODIA_TOKEN=your-token-here
curl --fail-with-body --silent --show-error \
-H "Authorization: Bearer $PRODIA_TOKEN" \
-H 'Accept: video/mp4' \
--json '{
"type": "inference.minimax.h3.fast.txt2vid.v1",
"config": {
"prompt": "an aerial shot drifting over a mountain ridge at sunrise, clouds rolling through the valley",
"duration": 6
}
}' \
--output video.mp4 \
https://inference.prodia.com/v2/job// npm install prodia
const fs = require("node:fs/promises");
const { createProdia } = require("prodia/v2");
const prodia = createProdia({ token: process.env.PRODIA_TOKEN });
(async () => {
const job = await prodia.job({
type: "inference.minimax.h3.fast.txt2vid.v1",
config: {
prompt: "an aerial shot drifting over a mountain ridge at sunrise, clouds rolling through the valley",
duration: 6,
},
});
await fs.writeFile("video.mp4", new Uint8Array(await job.arrayBuffer()));
})();Based on the H3 Fast API reference. A 6-second H3 Fast clip is $0.2793 at published rates.
We optimize the inference.
You build the product.
Our team builds implementations of models across different hardware, looking for the best balance of cost, quality and speed for each one. About Prodia
Tuned for the hardware
We run on NVIDIA GPUs, and on other AI chips too. Our Wan 2.2 implementation on Tenstorrent hardware is one example.
Research we publish
Our write-ups explain the methods, such as AnyFlow, a single LoRA that runs FLUX at any step count.
Prices you can check
Every model has a published per-job price, and the API can return the actual cost of a completed job.
Questions.
How do I get an API key?
Sign up at app.prodia.com, then create a token in the API dashboard. API tokens require a Pro subscription. The quickstart walks through your first request.
Where are prices listed?
Per-job prices for every model, by duration and resolution, are on the pricing page in the docs. For large workloads, use the volume pricing calculator.
Do I have to manage GPUs?
No. You send HTTPS requests to the Prodia API and we run the models.
Can I use Prodia with an AI coding assistant?
Yes. Point it at our agent guide or llms.txt. The documentation index is also available as Markdown.