Seedance 2.0 Fast Tier vs Standard: The Pricing Math

Seedance 2.0 has two tiers on fal: Standard and Fast. Same prompt surface, same 720p cap, same 4 to 15 second range. Fast trades some fine detail and narrative coherence for about 20% off the bill. This post runs the actual math.

The per second rates, spelled out

Text to video Standard: $0.3034 per second. Image to video Standard: $0.3024. Reference to video with image inputs: $0.3024. Reference to video with video inputs applies a 0.6 multiplier and lands at $0.1814 per second.

Fast tier text to video: $0.2419 per second. Fast reference to video with video input: $0.14515 per second. Fast drops token cost from $0.014 per 1K to $0.0112 per 1K.

The token formula you can put in a spreadsheet

Every Seedance 2.0 call bills tokens based on output dimensions, duration, and frame rate:

01tokens = (height x width x duration_seconds x 24) / 1024

At 720p (1280 x 720), 5 seconds, 24fps:

01(720 x 1280 x 5 x 24) / 1024 = 108,000 tokens

Standard: 108,000 x $0.014 / 1000 = $1.51. Per second rate ($0.3034 x 5 = $1.52) tracks the token math within rounding. Stick with per second for quick math.

The 100 clip budget table

100 ad variants, each 5 seconds at 720p:

Bump duration to 8 seconds: Standard lands at $242.72 for 100 clips, Fast at $193.52, saving $49.20.

A real production workflow with 3 to 5 iterations per final clip: 400 renders at 8 seconds is $970.88 Standard vs $774.08 Fast. The $196 difference is real money for a small studio.

Where Fast wins

1. Early concept exploration. You don't know if the shot works yet. You want 20 angles cheap.
2. Client reviews. Rough cut quality is fine for approval.
3. Prompt sweeps. Varying wording across 50 calls to find the right phrase.
4. Ad iteration. Testing 10 variants of the same 5 second hook on paid media, then re rendering the winner on Standard.
5. Social drafts. TikTok and Reels compress hard enough that Fast often looks the same at playback.

Where Fast loses

1. Fine texture work. Fabric weave, hair strand detail, reflective surfaces. Fast smooths these out.
2. Long narrative continuity. 12 to 15 second single shots with multiple actions. Fast drifts on the second half.
3. Complex physics. Water interactions, shattering glass, cloth simulation. Fast wobbles.
4. Multi speaker dialogue with tight lip sync. Fast's audio is less polished.
5. Hero shots for final delivery. Homepage or broadcast: render the winner at Standard.

The minimal Fast call

01import { fal } from "@fal-ai/client";
02
03const result = await fal.subscribe("bytedance/seedance-2.0/fast/text-to-video", {
04  input: {
05    prompt: "Aerial shot over a coastal highway at sunset, one red convertible cutting through switchback curves, golden hour side light.",
06    resolution: "720p",
07    duration: 6,
08    aspect_ratio: "16:9",
09    generate_audio: true
10  },
11  logs: true
12});
13
14console.log(result.data.video.url);
15// 6 seconds at $0.2419 = $1.45 per clip

Same prompt on Standard lands at $1.82. For a driving shot at 6 seconds you probably cannot tell the difference on a phone screen. On a 4K desktop, side by side, Standard holds reflection detail better.

The iteration workflow most shops converge on

1. Start every concept on Fast. Burn 10 to 20 calls to find the prompt.
2. Once the prompt is locked, render 3 to 5 variations on Standard to pick a hero.
3. If the hero has a motion issue, fix the prompt and re render on Standard only for the winner.

That pattern lands a finished clip for about $6 to $10 all in, versus $15 to $25 going Standard from the first prompt. Savings compound across a slate.