ByteDance Slashes Seedance 2.0 Mini Pricing by 50%, Igniting B2B Price War in Video Generation
N.R. Finch
ByteDance's Volcano Engine launched Seedance 2.0 Mini, a video-generation model priced at roughly ¥0.5 per second of 720p output — half the standard version's cost — signaling that the B2B AI-video battle has shifted from dazzling demos to hard unit economics.
¥0.5 per second — how cheap is that, really?
Seedance 2.0 Mini prices image-to-video at ¥0.023 per thousand tokens and video-to-video at ¥0.014 per thousand tokens. At 720p, that works out to roughly ¥0.5 per second.
In plain terms = one minute of 720p video costs about ¥30 in raw model calls. A 100-minute short drama using the model for filler shots, effects, or background renders could hit an extremely low theoretical cost.
This means → it undercuts the entire pricing logic of traditional stock-footage licensing and outsourced production — not a marginal saving, but a different cost structure altogether.
Standard vs. Mini — what is ByteDance's play?
Volcano Engine is running a tiered pricing strategy: the standard version targets premium film and TV production with high fidelity and complex semantic understanding; Mini captures the long-tail market and high-concurrency creative workflows on price.
The two versions price image-to-video and video-to-video separately and explicitly. This reflects that leading model providers can now precisely cost-account the compute each modality conversion — image-to-video vs. video-to-video — actually consumes.
Put simply = expensive version for high-end jobs, cheap version for bulk work — two lines, each earning its own margin.
Opening the API — what is ByteDance actually selling?
Mini's headline feature is an open API — a technical gateway letting outside developers and enterprises call the model directly — revealing a deeper commercial intent: MaaS, Model-as-a-Service.
This means → ByteDance is not just selling a video-generation tool. It aims to become foundational infrastructure, plugging into third-party apps, toolchains, and enterprise production pipelines, using developer-ecosystem stickiness to drive sustained cloud-compute consumption.
In plain terms = a tool collects one-time fees; infrastructure collects ongoing tolls.
What do B2B customers care about right now?
The low-price-plus-open-API combo has one core purpose: lowering the cost of experimentation for enterprise clients.
B2B buyers' focus has already shifted from "how impressive is the AI video output" to "can AI's marginal-cost reduction actually cover integration costs and trial-and-error costs."
This signals that AI-video commercialization is moving from the concept phase into the hard-math phase — enterprises are running the numbers, not just watching demos.
Content is for reference only, not financial advice.