Veo 3.1 Fast at $0.12 a second vs Kling 3 Pro at $0.14 or $0.21

Two popular tiers, side by side on price. Google's Gemini API pricing page (read 2026-10-03) lists Veo 3.1 Fast at $0.12 per second at 1080p. Sume lists Kling 3 Pro, billed at provider list times 1.25: $0.14 per second with audio off and $0.21 with audio on. Sume's Video API docs says audio defaults to the model's capability, so unless you pass generate_audio: false you pay the higher rate.

This is a price table, not a verdict on quality, and I do not state the resolution Kling 3 Pro renders at here; check its catalog entry first.

For 5, 8 and 10 seconds

Multiply each rate by the clip length.

What the gaps mean

Kling 3 Pro without audio is about 17 percent above Veo 3.1 Fast's list; with audio it is 75 percent above. If your clips will get a separate voice or music track, turning audio off saves a third of the Kling price: $0.07 per second, or $0.70 on a 10-second clip.

The trap is the default. A pipeline that does not set generate_audio pays the audio-on rate. If you add narration later, you have paid for audio you replace. Set the flag deliberately.

Setting the flag

A minimal request body on Sume's video endpoint that turns audio off and so prices at $0.14 per second looks like this. It only builds and prints the body; sending it needs your key.

import json

body = {
    'model': 'kling-3',
    'prompt': 'A ceramic mug on a wooden table, slow push-in',
    'duration': 8,
    'generate_audio': False,
}
print(json.dumps(body, indent=2))
print('expected cost: $%.2f' % (0.14 * body['duration']))