Short answer: Pick Veo 3.1 Lite for volume, tight budgets, or 1080p without the premium: 720p with sound is $0.10 per second, 40% of Fast's $0.25. Pick Veo 3.1 Fast when you need 4K or steadier audio rhythm. In our same-prompt test, Lite actually followed the prompt better: it showed the drummer's face and a train pulling in, while Fast cropped the face out and added a second drummer, though its drumbeat stayed regular throughout. Both are live on HiAPI: try Veo 3.1 Lite on its model page.
The differences at a glance
| Veo 3.1 Fast | Veo 3.1 Lite | |
|---|---|---|
| Model id | veo-3.1-fast/text-to-video, /image-to-video | veo-3.1-lite/text-to-video, /image-to-video |
| 720p / second | $0.25 with sound, $0.17 without | $0.10 with sound, $0.06 without |
| 1080p / second | $0.25 with sound, $0.17 without | $0.14 with sound, $0.10 without |
| 4K | Yes, $0.60 with sound, $0.50 without | No |
| Duration | 4, 6 or 8 s | 4, 6 or 8 s |
| Aspect ratio | 16:9, 9:16 | 16:9, 9:16 |
| Sound | generate_audio, on by default | generate_audio |
| Time for a 4 s clip in this test | 166 s | 107 s |
Live HiAPI prices, checked October 18, 2026. Times are from one run each.
Fast charges the same for 720p and 1080p; Lite's 1080p costs more than its 720p. An 8-second 1080p clip with sound is $2.00 on Fast and $1.12 on Lite.
Same prompt, side by side
Both at 720p, 16:9, 4 seconds, with sound.
A street drummer plays on upturned plastic buckets in a busy subway station at rush hour, close shot on the sticks blurring and his focused face, commuters streaming past behind him, a train pulling in on the right, handheld camera slowly circling. Sound: crisp bucket drumming, a train braking, crowd footsteps and a station announcement echoing.
Veo 3.1 Fast:
veo-3.1-fast/text-to-video, 720p, 4 seconds, with sound, $1.00.
Veo 3.1 Lite:
veo-3.1-lite/text-to-video, 720p, 4 seconds, with sound, $0.40.
What's visibly different:
- Prompt adherence: the prompt asked for "the sticks blurring and his focused face." Lite shows the whole drummer, face, hands and buckets. Fast frames so tight that the face is cut off, and adds a second drummer in the background wearing a marching-drum harness, so it reads more like a drumline.
- The train: Lite's train arrives from the distance with its headlights on, matching "pulling in." Fast's train stays parked on the right.
- Camera: the prompt asked for a handheld orbit. Neither really moves; Fast has only a slight drift.
- Sound: both clips have an audio track. On the volume waveform, Fast's drum hits land regularly, about one every 0.35 seconds; Lite's drumming thins out noticeably between 1.7 and 3 seconds.
One test can't settle everything, but it does show that paying 2.5× for Fast doesn't guarantee a picture that follows the prompt more closely.
Which to choose
- Batch production, budget first: Lite. 100 clips of 8 seconds at 720p with sound cost $80 on Lite and $200 on Fast.
- 1080p: Lite, $0.14 per second with sound, versus $0.25 on Fast.
- 4K: Fast only, $0.60 per second with sound.
- No sound needed: both let you turn off
generate_audio. Lite drops to $0.06 per second at 720p, Fast to $0.17. - Steady audio rhythm (music, beats, voiceover): Fast held it better here; run a 4-second test on both before committing.
- The best picture quality: look at standard Veo 3.1, $0.57 per second at 720p/1080p with sound; see Veo 3.1 API pricing.
Switching between them
The schemas are nearly identical, so you mostly just change model. Two things to watch: Lite rejects 4k, and Fast requires resolution and duration.
{"model": "veo-3.1-fast/text-to-video", "input": {"prompt": "...", "resolution": "720p", "duration": 4, "aspect_ratio": "16:9", "generate_audio": true}}
{"model": "veo-3.1-lite/text-to-video", "input": {"prompt": "...", "resolution": "720p", "duration": 4, "aspect_ratio": "16:9", "generate_audio": true}}
duration accepts only the integers 4, 6 and 8; anything else returns HTTP 400.
FAQ
Is Lite lower quality than Fast?
Not visibly in this test, and Lite followed the prompt more closely. Both came back at 1280×720. The real differences are price, 4K and audio rhythm.
Which one is faster?
Here Lite returned in 107 seconds and Fast in 166. Speed varies with load, so time your own prompts.
Which should I use to test prompts?
Lite at 4 seconds, 720p, with sound: $0.40 per try. Once the prompt works, decide whether to move up to Fast or standard.
Do they do image-to-video?
Yes. Both have an /image-to-video endpoint at the same price as text-to-video.
Full prices for all three Veo 3.1 tiers are in Veo 3.1 API pricing, and curl and Python code is in the Veo 3.1 API guide. Or copy the prompt above and generate it on Lite for $0.40.
