AI video models are usually priced per second of generated video, and prices vary by more than ten times between models and resolutions. This guide collects official list prices from each developer's own pricing page as of October 9, 2026, so you can estimate what a clip will cost before you generate it.
Important: these are API list prices. Apps and platforms (including TextVidAI) set their own prices or credit costs, and providers change prices often. Always check the source links before budgeting.
Price per second by model (official list prices)
| Model | Resolution | Price per second (USD) | Source |
|---|---|---|---|
| Google Veo 3.1 Standard | 720p / 1080p | $0.40 (4K: $0.60) | |
| Google Veo 3.1 Fast | 720p / 1080p / 4K | $0.10 / $0.12 / $0.30 | |
| Google Veo 3.1 Lite | 720p / 1080p | $0.05 / $0.08 (no 4K) | |
| Gemini Omni Flash | 720p | about $0.10 (token-based) | |
| Wan 3.0 | 480p / 720p / 1080p | $0.05 / $0.10 / $0.20 | Alibaba Cloud |
| MiniMax H3 | 768p / 2K | $0.08 / $0.13 | MiniMax |
| MiniMax H3-Max | 480p / 768p | $0.05 / $0.08 | MiniMax |
| Vidu Q4 Preview | 540p to 4K | from $0.045 up to $0.39 | Shengshu press release |
| Runway Gen-4.5 | standard | $0.12 (12 credits at $0.01) | Runway |
Veo 3.1 prices include audio. Gemini Omni Flash is billed per token; Google converts this to roughly $0.10 per second at 720p.
Models priced per clip
Luma Ray3.2 is priced per clip rather than per second. Standard clips cost $0.30 (5 s) or $0.90 (10 s) at 720p, and $1.20 (5 s) or $3.60 (10 s) at 1080p; draft quality at 360p starts at $0.06. Luma notes that rates may change before general availability (Luma pricing).
Models without a verifiable public per-second price
- Kling 3.0 / 4.0: Kling's API pricing page did not show prices we could verify. Check Kling's API pricing directly.
- Seedance 2.5: ByteDance's BytePlus platform prices Seedance by tokens. Through Runway's API, Seedance 2.5 costs 20, 30 or 68 credits per second at 480p, 720p or 1080p ($0.20, $0.30 or $0.68 at $0.01 per credit).
What a typical clip costs
- 8 seconds at 1080p: Veo 3.1 Standard $3.20, Veo 3.1 Fast $0.96, Veo 3.1 Lite $0.64, Wan 3.0 $1.60.
- 8 seconds on MiniMax H3 at 768p: $0.64.
- 10 seconds on Runway Gen-4.5: $1.20.
- 30 seconds on Wan 3.0: $1.50 at 480p, $6.00 at 1080p.
How to spend less on AI video
- Draft cheap, finish expensive. Test prompts on a fast or low-resolution tier (Veo 3.1 Lite or Fast, Wan 3.0 at 480p), then render the final at higher quality.
- Keep test clips short. A 4-second test costs half of an 8-second one.
- Only pay for 4K when you need it. 4K often costs 1.5–3 times more than 1080p.
- Fix the prompt before regenerating. Most wasted spend is repeated generations with small, random changes. See our prompt ideas and camera movement prompts.
- Use one credit balance across models instead of several subscriptions.
What it costs on TextVidAI
TextVidAI uses credits rather than per-second billing, with a one-time starter pack and subscription plans. See the pricing page for current credit costs per model.
FAQ
What is the cheapest AI video model per second?
Among the official prices above, the lowest are Vidu Q4 Preview at 540p ($0.045/s), Veo 3.1 Lite at 720p, Wan 3.0 at 480p and MiniMax H3-Max at 480p (all $0.05/s).
Does Veo 3.1 cost extra for audio?
No separate charge: Google's listed Veo 3.1 prices are for video with audio.
How often do these prices change?
Often. Treat this page as a snapshot dated October 9, 2026 and check each source before you budget.
Prices checked: October 9, 2026.