Gemini Omni Flash cost · Runway API partner models · ClipCost · AI video API pricing · Cheapest AI video generator · Wan API pricing · Kling API pricing · LTX API pricing · LTX vs Wan cost · Veo 3 cost · Published Oct 25, 2026 · Updated Oct 27, 2026 (9am: Grok Imagine 1.5 / 1.5 Lite added, new #1; 3am: Runway API Seedance 2.0 price cut re-ranked) · Grok Imagine cost · Rates checked Oct 19–27, 2026 · Runway API pricing · Vidu cost · Prices from verified data (JSON)
Cheapest AI video API: every API ranked by what a usable clip costs
If you generate video from code, you pay per second of output — and the cheapest second depends on what you need: sound or silence, a start image or just a prompt, 720p or 4K. Here are the 166 pay-per-second API rates ClipCost has verified across eight APIs, cut by use case, by provider and by “same model, cheaper route”, with what a usable 30-second clip really costs once you allow for retries.
Run your own numbers Compare tools Full API price list
Quick answer
The cheapest AI video API is Grok Imagine Video 1.5 Lite at 480p: $0.02 per second on xAI’s own API or the Runway API — $0.60 per 30-second try and $1.80 per usable 30-second clip with 2 retries, native audio included, text- or image-to-video, 1–15s per clip (Grok Imagine cost). Next is Wan 2.6 Flash at 720p without audio on Alibaba Cloud Model Studio: $0.025 per second ($2.25 usable 30s), image-to-video only, tied since Oct 31, 2026 with MiniMax H3-Lite 480P on MiniMax’s own API (MiniMax Hailuo cost). At 720p for text-to-video, Veo 3.1 Lite without audio on Google Cloud Vertex AI is $0.03/sec ($2.70 usable 30s), tied with LTX-2.3 Fast 720p on LTX’s own API and Grok Imagine 1.5 Lite 720p (both with audio), then Kling 2.6 / 2.5 Turbo 720p on Kling’s own API ($0.042, $3.78); Vidu Q3 Turbo 540p on Vidu’s API is $0.035 ($3.15, text- or image-to-video, audio on by default; Vidu cost) and Seedance 2.0 Mini 480p on BytePlus is $0.036 ($3.24) but its input modes are not stated on the price page. With native audio the floor is now $0.02/sec ($1.80 usable 30s): Grok Imagine 1.5 Lite 480p; at 720p it is $0.03 ($2.70) on Grok Imagine 1.5 Lite or LTX-2.3 Fast on LTX’s own API (audio generated by default, not priced separately — LTX API pricing), then Vidu Q3 Turbo 540p $0.035 ($3.15), LTX-2.3 Pro 720p $0.04 ($3.60) and $0.05 ($4.50) on Veo 3.1 Lite 720p, Wan 2.6 Flash 720p or Wan 3.0 480p on the Runway API. The cheapest 1080p is Wan 2.6 Flash at $0.0375 silent ($3.38); with audio it is LTX-2.3 Fast at $0.06 ($5.40), then Wan 2.6 Flash $0.075 ($6.75) and Veo 3.1 Lite $0.08. The cheapest 4K is LTX-2.3 Fast at $0.24 per second with audio ($21.60 usable 30s); Veo 3.1 Fast is listed at $0.25 / $0.30 ⚑. Going direct to the model maker is never dearer than the Runway API in our data, and is cheaper for most models (Grok Imagine 1.5 Lite, Wan 3.0, Wan 3.0 Prime and Veo 3.1 Standard cost the same either way).
New Nov 1, 2026 (9am): 13 Vidu API rates join the API ranking (now 138 rates), verified today on Vidu’s API pricing page (1 credit = $0.005) and API reference: Q3 Turbo $0.035 / $0.06 / $0.065 and Q3 Pro $0.045 / $0.10 / $0.12 per second at 540p / 720p / 1080p, Q3 Pro Fast $0.06 / $0.075 (720p / 1080p, image-to-video only) and Q4 Preview $0.045–$0.39 (540p to 4K, image/reference-to-video only), audio on by default. Q3 Turbo 540p ($1.05 per 30s try, $3.15 usable) is API #9. Full breakdown: Vidu cost.
New Oct 31, 2026 (9pm): four MiniMax rates join the API ranking (now 125 rates), re-verified today on MiniMax’s video pricing page and API reference: H3-Lite $0.025 / $0.04 and H3-Highspeed $0.04 / $0.06 per second at 480P / 768P (4–15s clips, no 2K; native audio not stated). H3-Lite 480P ($0.75 per 30s try, $2.25 usable) ties Wan 2.6 Flash 720p silent for #3=, behind only Grok Imagine 1.5 Lite 480p. Full breakdown: MiniMax Hailuo cost.
New #1, Oct 27, 2026 (9am): xAI’s Grok Imagine Video 1.5 Lite is $0.02 / $0.03 / $0.14 per second at 480p / 720p / 1080p (1080p is 720p upscaled) with native audio on xAI’s own API, verified today on xAI API pricing, and the same on the Runway API (2 / 3 / 14 credits per second). 480p is $0.60 per 30-second try and $1.80 per usable 30s — the cheapest of all 166 API rates and all 210 rates ClipCost ranks, ahead of Wan 2.6 Flash 720p silent ($2.25). The full Grok Imagine 1.5 is $0.08 / $0.14 / $0.25 on xAI and $0.10 / $0.16 / $0.29 on the Runway API. Clips are 1–15s, so a 30-second video takes at least two generations. See Grok Imagine cost.
Updated Oct 27, 2026 (3am): Runway cut its API price for Seedance 2.0 and now bills it by resolution (re-checked today on Runway API pricing): seedance2_mini 7 / 15, seedance2_fast 11 / 24 and seedance2 14 / 30 / 40 / 150 credits per second at 480p / 720p / 1080p / 4K ($0.01 per credit; was 16, 29 and 36–150). The cheapest is Seedance 2.0 Mini 480p at $0.07 per second ($6.30 per usable 30s) — still about double BytePlus ($0.036), so the ranking gained 5 Runway rows (109 rates at 3am) but no new leader. Full list: Runway API pricing.
Ranked Oct 26, 2026 (3pm): LTX’s own API joins the ranking with 16 rates re-verified today on docs.ltx.io/pricing. LTX-2.3 Fast at 720p is $0.03 per second with audio generated by default ($0.90 per 30-second try, $2.70 usable) — then tied #2 with Veo 3.1 Lite 720p without audio and the cheapest API with native audio until Oct 27 (was $0.05). LTX-2.3 Fast is also the cheapest 1080p with audio ($0.06) and the cheapest 4K ($0.24). See LTX API pricing and LTX vs Wan cost.
Cheapest per usable 30 seconds: APIs hold seven of the nine cheapest spots among all 210 rates on ClipCost (Grok Imagine 1.5 Lite 480p $1.80; Wan 2.6 Flash 720p and MiniMax H3-Lite 480P $2.25; Veo 3.1 Lite, Luma Ray3.2 540p API, LTX-2.3 Fast 720p and Grok Imagine 1.5 Lite 720p $2.70); the cheapest subscription app, Luma Ray3.2 540p on Luma Plus annual credits, ties for second at $2.25 (since Nov 2, 2026; Luma cost) and Kling 2.6 Standard mode ties at $2.70 — only if you use every credit. See the cheapest AI video generator ranking for apps and APIs together.
Basis: list prices in US dollars per second of output video from each provider’s official page — Alibaba Cloud Model Studio pricing (Singapore / International, checked Oct 24, 2026), Google Cloud Agent Platform (Vertex AI) pricing (Oct 20, 2026; re-checked Oct 22), BytePlus ModelArk pricing (Oct 22–23, 2026), Kling API video pricing (1 unit = $0.14, Oct 25, 2026), LTX API pricing (Oct 26, 2026), xAI API pricing (Oct 27, 2026), MiniMax video pricing (Oct 19, 2026; re-checked Oct 22) and Runway API pricing (credits per second × $0.01, Oct 22–25, 2026; Seedance 2.0 and Grok Imagine checked Oct 27, 2026). The same 138 rates, in the same order, are listed on AI video API pricing. “Usable” = 3 attempts (1 try + 2 retries), the same maths as our calculator. Ties share a rank. Excludes talking-presenter APIs (Hedra), editing (Aleph 2.0) and estimates (Hailuo 2.3 Fast). Prices exclude tax, input media and storage; no discounts are applied. Cost only — no quality ranking. Some links on ClipCost are affiliate links (see footer); they never change these numbers.
Cheapest AI video API by use case
| You need | Cheapest API & model | $ / sec | 30s try | 30s usable | Runner-up |
|---|---|---|---|---|---|
| Cheapest overall | xAI API: Grok Imagine 1.5 Lite · 480p, audio on Runway API: Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) | $0.02 | $0.60 | $1.80 | Alibaba Cloud Model Studio Wan 2.6 Flash · 720p, no audio, image-to-video — $0.025 ($2.25), tied with MiniMax API H3-Lite · 480P (Oct 31, 2026) |
| Text-to-video (no start image needed) | xAI API: Grok Imagine 1.5 Lite · 480p, audio on Runway API: Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) | $0.02 | $0.60 | $1.80 | Google Vertex AI Veo 3.1 Lite · 720p, no audio — $0.03 ($2.70) |
| With native audio (stated by the provider) | xAI API: Grok Imagine 1.5 Lite · 480p, audio on Runway API: Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) | $0.02 | $0.60 | $1.80 | LTX API LTX-2.3 Fast · 720p, audio on — $0.03 ($2.70) |
| Native audio, text-to-video | xAI API: Grok Imagine 1.5 Lite · 480p, audio on Runway API: Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) | $0.02 | $0.60 | $1.80 | LTX API LTX-2.3 Fast · 720p, audio on — $0.03 ($2.70) |
| 480p | xAI API: Grok Imagine 1.5 Lite · 480p, audio on Runway API: Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) | $0.02 | $0.60 | $1.80 | MiniMax API H3-Lite · 480P — $0.025 ($2.25) |
| 720p, silent | Alibaba Cloud Model Studio: Wan 2.6 Flash · 720p, no audio, image-to-video | $0.025 | $0.75 | $2.25 | Google Vertex AI Veo 3.1 Lite · 720p, no audio — $0.03 ($2.70) |
| 720p, silent, text-to-video | Google Vertex AI: Veo 3.1 Lite · 720p, no audio | $0.03 | $0.90 | $2.70 | Kling API Kling 2.6 · 720p, no native audio — $0.042 ($3.78) |
| 1080p, silent | Alibaba Cloud Model Studio: Wan 2.6 Flash · 1080p, no audio, image-to-video | $0.0375 | $1.13 | $3.38 | Google Vertex AI Veo 3.1 Lite · 1080p, no audio — $0.05 ($4.50) |
| 1080p, silent, text-to-video | Google Vertex AI: Veo 3.1 Lite · 1080p, no audio | $0.05 | $1.50 | $4.50 | Kling API Kling 2.6 · 1080p, no native audio — $0.07 ($6.30) |
| 1080p with native audio | LTX API: LTX-2.3 Fast · 1080p, audio on | $0.06 | $1.80 | $5.40 | Alibaba Cloud Model Studio Wan 2.6 Flash · 1080p, with audio, image-to-video — $0.075 ($6.75) |
| 1080p with native audio, text-to-video | LTX API: LTX-2.3 Fast · 1080p, audio on | $0.06 | $1.80 | $5.40 | Google Vertex AI Veo 3.1 Lite · 1080p with audio — $0.08 ($7.20) |
| 4K (any audio) | LTX API: LTX-2.3 Fast · 4K, audio on ⚑ | $0.24 | $7.20 | $21.60 | Google Vertex AI Veo 3.1 Fast · 4K, no audio — $0.25 ($22.50) |
| 4K with native audio | LTX API: LTX-2.3 Fast · 4K, audio on ⚑ | $0.24 | $7.20 | $21.60 | Google Vertex AI Veo 3.1 Fast · 4K with audio — $0.30 ($27.00) |
| Long single shots (up to 30s per generation) | Alibaba Cloud Model Studio: Wan 3.0 · 480p | $0.05 | $1.50 | $4.50 | Alibaba Cloud Model Studio Wan 3.0 Prime (fast) · 480p — $0.068 ($6.12) |
| Kling models | Kling API: Kling 2.6 · 720p, no native audio Kling API: Kling 2.5 Turbo · 720p, no native audio | $0.042 | $1.26 | $3.78 | Kling API Kling 2.6 · 1080p, no native audio — $0.07 ($6.30) |
| LTX models | LTX API: LTX-2.3 Fast · 720p, audio on | $0.03 | $0.90 | $2.70 | LTX API LTX-2.3 Pro · 720p, audio on — $0.04 ($3.60) |
| Grok Imagine (xAI) models | xAI API: Grok Imagine 1.5 Lite · 480p, audio on | $0.02 | $0.60 | $1.80 | xAI API Grok Imagine 1.5 Lite · 720p, audio on — $0.03 ($2.70) |
“Native audio” counts only rates where the provider says the output has audio; MiniMax, BytePlus Seedance, Wan 3.0 / HappyHorse on Alibaba Cloud and some Runway API models do not state it, so they are left out of the audio rows. LTX generates audio by default at no extra charge, so its rows count as native audio. Wan 2.6 Flash and Runway Gen-4 Turbo need a start image. xAI states native audio for both Grok Imagine 1.5 models; Runway’s API docs do not mention audio, so the Runway API Grok rows use xAI’s statement. “Text-to-video” rows only count rates whose text-to-video support we have checked on an official page (Google’s Veo 3.1 model page, Kling’s API capability map, Alibaba’s Wan 3.0 / HappyHorse notes); “Check docs” in the Input column means the price page does not say. ⚑ 4K: Google’s Vertex AI price list includes Veo 3.1 Fast at 4K, but its Veo 3.1 model page (checked Oct 25, 2026) lists only 720p and 1080p output for Veo 3.1 Fast (4K for Veo 3.1 Standard) — confirm before relying on the $0.25 / $0.30 4K rate. Kling 2.6 audio ⚑: Kling’s price list charges $0.14/sec at 1080p while its capability map says Kling 2.6 native audio is not supported — confirm before you build on it.
The 26 cheapest AI video API rates (ties included)
Per try, per usable clip with 2 retries, and how many usable seconds $10 buys. The other 120 rates (up to $1.50 per second) are on AI video API pricing.
| # | API | Model | Res. | Native audio | Input | Length per generation | $ / sec | 30s try | 30s usable | 60s usable | Usable seconds per $10 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1= | xAI API | Grok Imagine 1.5 Lite · 480p, audio on | 480p | Yes | Text or image | 1–15s | $0.02 | $0.60 | $1.80 | $3.60 | 167s |
| 1= | Runway API | Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) | 480p | Yes | Text or image | 1–15s | $0.02 | $0.60 | $1.80 | $3.60 | 167s |
| 3= | Alibaba Cloud Model Studio | Wan 2.6 Flash · 720p, no audio, image-to-video | 720p | No | Image-to-video only | Up to 15s | $0.025 | $0.75 | $2.25 | $4.50 | 133s |
| 3= | MiniMax API | MiniMax H3-Lite API · 480P New Oct 31 | 480p | Not stated | Check docs | 4–15s | $0.025 | $0.75 | $2.25 | $4.50 | 133s |
| 5= | Google Vertex AI | Veo 3.1 Lite · 720p, no audio | 720p | No | Text or image | 4, 6 or 8s | $0.03 | $0.90 | $2.70 | $5.40 | 111s |
| 5= | LTX API | LTX-2.3 Fast · 720p, audio on | 720p | Yes | Text or image | 6–20s | $0.03 | $0.90 | $2.70 | $5.40 | 111s |
| 5= | xAI API | Grok Imagine 1.5 Lite · 720p, audio on | 720p | Yes | Text or image | 1–15s | $0.03 | $0.90 | $2.70 | $5.40 | 111s |
| 5= | Runway API | Grok Imagine 1.5 Lite · 720p grok_imagine_1_5_lite (720p) | 720p | Yes | Text or image | 1–15s | $0.03 | $0.90 | $2.70 | $5.40 | 111s |
| 5= | Luma API | Luma Ray3.2 · 540p New Nov 2 | 540p | No | Text or image | 5s or 10s (10s = 3× the 5s price) | $0.03 | $0.90 | $2.70 | $5.40 | 111s |
| 10 | Vidu API | Vidu Q3 Turbo · 540p, audio on | 540p | Yes | Text or image | 1–16s | $0.035 | $1.05 | $3.15 | $6.30 | 95s |
| 11 | BytePlus ModelArk | Seedance 2.0 Mini API · 480p (no video input) | 480p | Not stated | Check docs | 4–15s | $0.036 | $1.08 | $3.24 | $6.48 | 93s |
| 12 | Alibaba Cloud Model Studio | Wan 2.6 Flash · 1080p, no audio, image-to-video | 1080p | No | Image-to-video only | Up to 15s | $0.0375 | $1.13 | $3.38 | $6.75 | 89s |
| 13= | LTX API | LTX-2.3 Pro · 720p, audio on | 720p | Yes | Text or image | 6, 8 or 10s | $0.04 | $1.20 | $3.60 | $7.20 | 83s |
| 13= | MiniMax API | MiniMax H3-Lite API · 768P New Oct 31 | 768p | Not stated | Check docs | 4–15s | $0.04 | $1.20 | $3.60 | $7.20 | 83s |
| 13= | MiniMax API | MiniMax H3-Highspeed API · 480P New Oct 31 | 480p | Not stated | Check docs | 4–15s | $0.04 | $1.20 | $3.60 | $7.20 | 83s |
| 16= | Kling API | Kling 2.6 · 720p, no native audio | 720p | No | Text or image | 5s or 10s | $0.042 | $1.26 | $3.78 | $7.56 | 79s |
| 16= | Kling API | Kling 2.5 Turbo · 720p, no native audio | 720p | No | Text or image | 5s or 10s | $0.042 | $1.26 | $3.78 | $7.56 | 79s |
| 18= | Vidu API | Vidu Q3 Pro · 540p, audio on | 540p | Yes | Text or image | 1–16s | $0.045 | $1.35 | $4.05 | $8.10 | 74s |
| 18= | Vidu API | Vidu Q4 Preview · 540p, audio on | 540p | Yes | Image or reference | 3–16s | $0.045 | $1.35 | $4.05 | $8.10 | 74s |
| 20= | Google Vertex AI | Veo 3.1 Lite · 720p with audio | 720p | Yes | Text or image | 4, 6 or 8s | $0.05 | $1.50 | $4.50 | $9.00 | 67s |
| 20= | Google Vertex AI | Veo 3.1 Lite · 1080p, no audio | 1080p | No | Text or image | 4, 6 or 8s | $0.05 | $1.50 | $4.50 | $9.00 | 67s |
| 20= | Alibaba Cloud Model Studio | Wan 3.0 · 480p | 480p | Not stated | Text, image or reference | 2–30s | $0.05 | $1.50 | $4.50 | $9.00 | 67s |
| 20= | Alibaba Cloud Model Studio | Wan 2.6 Flash · 720p, with audio, image-to-video | 720p | Yes | Image-to-video only | Up to 15s | $0.05 | $1.50 | $4.50 | $9.00 | 67s |
| 20= | MiniMax API | MiniMax H3-Max API · 480P | 480p | Not stated | Check docs | 5–15s | $0.05 | $1.50 | $4.50 | $9.00 | 67s |
| 20= | Runway API | Wan 3.0 · 480p wan3 (480p) | 480p | Yes | Check docs | See Runway docs | $0.05 | $1.50 | $4.50 | $9.00 | 67s |
| 20= | Runway API | Gen-4 Turbo gen4_turbo | Not stated | Not stated | Image-to-video only | See Runway docs | $0.05 | $1.50 | $4.50 | $9.00 | 67s |
Maths: $ per second × 30 = one 30s try; × 3 = usable. Usable seconds per $10 = $10 ÷ ($ per second × 3). Veo bills whole 4, 6 or 8-second clips; Seedance 2.0 on BytePlus is priced per 5-second video; Kling 2.6 / 2.5 Turbo make 5 or 10-second clips — a 30-second video is several generations stitched together at the same per-second price.
Cheapest rate on each API
| API | Rates | Cheapest (usable 30s) · rank of 146 | Cheapest with native audio | Cheapest 1080p | Most expensive |
|---|---|---|---|---|---|
| xAI API | 6 | Grok Imagine 1.5 Lite · 480p, audio on — $0.02 ($1.80) · #1= | Grok Imagine 1.5 Lite · 480p, audio on — $0.02 ($1.80) | Grok Imagine 1.5 Lite · 1080p, audio on — $0.14 ($12.60) | Grok Imagine 1.5 · 1080p, audio on — $0.25 ($22.50) |
| Black Forest Labs API | 5 | FLUX 3 Draft · HD, audio on — $0.06 ($5.40) · #30= | FLUX 3 Draft · HD, audio on — $0.06 ($5.40) | FLUX 3 · FHD, audio on — $0.24 ($21.60) | FLUX 3 · UHD, audio on — $0.41 ($36.90) |
| Alibaba Cloud Model Studio | 19 | Wan 2.6 Flash · 720p, no audio, image-to-video — $0.025 ($2.25) · #3= | Wan 2.6 Flash · 720p, with audio, image-to-video — $0.05 ($4.50) | Wan 2.6 Flash · 1080p, no audio, image-to-video — $0.0375 ($3.38) | Wan 3.0 Prime (fast) · 1080p — $0.28 ($25.20) |
| Google Vertex AI | 14 | Veo 3.1 Lite · 720p, no audio — $0.03 ($2.70) · #5= | Veo 3.1 Lite · 720p with audio — $0.05 ($4.50) | Veo 3.1 Lite · 1080p, no audio — $0.05 ($4.50) | Veo 3.1 Standard · 4K with audio — $0.60 ($54.00) |
| BytePlus ModelArk | 10 | Seedance 2.0 Mini API · 480p (no video input) — $0.036 ($3.24) · #11 | None stated | Seedance 2.0 API · 1080p (no video input) — $0.374 ($33.66) | Seedance 2.5 API · 1080p (no video input) — $0.569 ($51.17) |
| Kling API | 17 | Kling 2.6 · 720p, no native audio — $0.042 ($3.78) · #16= | Kling 3.0 Turbo · 720p, native audio — $0.112 ($10.08) | Kling 2.6 · 1080p, no native audio — $0.07 ($6.30) | Kling 3.0 Omni · 4K — $0.42 ($37.80) |
| LTX API | 16 | LTX-2.3 Fast · 720p, audio on — $0.03 ($2.70) · #5= | LTX-2.3 Fast · 720p, audio on — $0.03 ($2.70) | LTX-2.3 Fast · 1080p, audio on — $0.06 ($5.40) | LTX-2.5 Pro · 4K, audio on — $0.39 ($35.10) |
| MiniMax API | 8 | MiniMax H3-Lite API · 480P — $0.025 ($2.25) · #3= | None stated | None stated | MiniMax H3 API · 2K — $0.13 ($11.70) |
| Vidu API | 13 | Vidu Q3 Turbo · 540p, audio on — $0.035 ($3.15) · #10 | Vidu Q3 Turbo · 540p, audio on — $0.035 ($3.15) | Vidu Q3 Turbo · 1080p, audio on — $0.065 ($5.85) | Vidu Q4 Preview · 4K, audio on — $0.39 ($35.10) |
| Luma API | 3 | Luma Ray3.2 · 540p — $0.03 ($2.70) · #5= | None (Ray3.2 has no audio) | Luma Ray3.2 · 1080p — $0.24 ($21.60) | Luma Ray3.2 · 1080p — $0.24 ($21.60) |
| Runway API | 51 | Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) — $0.02 ($1.80) · #1= | Grok Imagine 1.5 Lite · 480p grok_imagine_1_5_lite (480p) — $0.02 ($1.80) | Grok Imagine 1.5 Lite · 1080p (720p upscaled) grok_imagine_1_5_lite (1080p) — $0.14 ($12.60) | Seedance 2.0 · 4K with audio seedance2 (4K) — $1.50 ($135.00) |
Billing differences that move the real price: Alibaba Cloud and Kling bill only successful generations (failed requests are free); Kling sells prepaid unit packages that expire, and package sizes and minimum purchase are not on its public page (unverified); LTX bills a prepaid dollar balance (minimum purchase and expiry not published) and makes 6–20-second clips, so there is no 5-second option; Veo 3.1 on Vertex AI needs a Google Cloud project with access and quota — the Veo 3.1 model page lists fixed quota and Provisioned Throughput, not standard pay-as-you-go; the Runway API has per-generation minimums on some models (Seedance 2.0 Mini 28 credits, Fast 44, Seedance 2.0 56, Seedance 2.5 80, Aleph 2.0 56); MiniMax bills extra reference images and input video. Details on each tool page: Alibaba Cloud, Veo, Seedance, Kling API, LTX API, MiniMax, Runway.
Same model, cheaper API
| Model | Direct from the model maker ($ / sec) | Runway API ($ / sec) | Cheaper |
|---|---|---|---|
| Wan 3.0 480p / 720p / 1080p | Alibaba Cloud Model Studio: $0.05 / $0.10 / $0.20 | Runway API: $0.05 / $0.10 / $0.20 | Same price |
| Wan 3.0 Prime 480p / 720p / 1080p | Alibaba Cloud Model Studio: $0.068 / $0.14 / $0.28 | Runway API: $0.068 / $0.14 / $0.28 | Same price |
| HappyHorse 1.0 720p / 1080p | Alibaba Cloud Model Studio: $0.14 / $0.24 | Runway API: $0.15 / $0.30 | Alibaba direct |
| Seedance 2.5 480p / 720p / 1080p | BytePlus ModelArk: $0.103 / $0.231 / $0.569 | Runway API: $0.20 / $0.30 / $0.68 | BytePlus direct |
| Seedance 2.0 Mini 480p / 720p | BytePlus ModelArk: $0.036 / $0.076 | Runway API: $0.07 / $0.15 | BytePlus direct |
| Seedance 2.0 Fast 480p / 720p | BytePlus ModelArk: $0.056 / $0.12 | Runway API: $0.11 / $0.24 | BytePlus direct |
| Seedance 2.0 480p / 720p / 1080p | BytePlus ModelArk: $0.07 / $0.152 / $0.374 | Runway API: $0.14 / $0.30 / $0.40 | BytePlus direct (close at 1080p) |
| Veo 3.1 Fast with audio 720p / 1080p | Google Vertex AI: $0.10 / $0.12 | Runway API: $0.15 | Vertex AI direct |
| Veo 3.1 Fast without audio 720p / 1080p | Google Vertex AI: $0.08 / $0.10 | Runway API: $0.10 | Vertex AI direct (or tie at 1080p) |
| Veo 3.1 with audio | Google Vertex AI: $0.40 | Runway API: $0.40 | Same price |
| Veo 3.1 without audio | Google Vertex AI: $0.20 | Runway API: $0.20 | Same price |
| Kling 3.0 Omni with audio 720p / 1080p / 4K vs Kling O3 (= Kling 3.0 Omni) Std / Pro / 4K | Kling API: $0.112 / $0.14 / $0.42 | Runway API: $0.13 / $0.17 / $0.43 | Kling API direct |
| Grok Imagine 1.5 Lite 480p / 720p / 1080p | xAI API: $0.02 / $0.03 / $0.14 | Runway API: $0.02 / $0.03 / $0.14 | Same price |
| Grok Imagine 1.5 480p / 720p / 1080p | xAI API: $0.08 / $0.14 / $0.25 | Runway API: $0.10 / $0.16 / $0.29 | xAI direct |
| Kling 2.5 Turbo 720p / 1080p vs Kling 2.5 Turbo Pro on Runway | Kling API: $0.042 / $0.07 | Runway API: $0.12 | Kling API direct |
Runway’s API resolution names do not always map one-to-one to the maker’s: Kling O3 Standard / Pro are treated as 720p / 1080p, Runway’s Seedance 2.0 Mini and Veo 3.1 Fast rates cover several resolutions at one price, and Kling 2.5 Turbo Pro on Runway is not the same tier as Kling 2.5 Turbo on Kling’s API. More: Runway models ranked by cost, Wan vs Kling cost, Seedance vs Veo cost, Kling O3 cost, HappyHorse cost.
How to pay less for an AI video API
- Skip native audio when you can — or use an API that includes it. Silent output is half price on Wan 2.6 Flash ($0.025 vs $0.05), 40% cheaper on Veo 3.1 Lite ($0.03 vs $0.05) and a third cheaper on Kling 3.0 ($0.084 vs $0.126). LTX-2.3 Fast includes generated audio at $0.03 per second.
- Start from an image — or from a prompt on Grok Imagine 1.5 Lite. Grok Imagine 1.5 Lite 480p ($0.02, text or image, with audio) is the cheapest rate on any API; Wan 2.6 Flash image-to-video ($0.025, silent) is next; a still from an image model usually costs cents.
- Draft low, finish high. Iterate at 480p / 720p (Seedance 2.0 Mini 480p $0.036, Wan 3.0 480p $0.05) and render only the keeper at 1080p.
- Go direct, not through an aggregator. The Runway API never beat the model maker’s own price in our table; it ties on Grok Imagine 1.5 Lite, Wan 3.0, Wan 3.0 Prime, Veo 3.1 Standard and Veo 3.1 Fast 1080p without audio.
- Cut retries, not seconds. Every regeneration is billed in full — tighter prompts and a locked first frame save more than shaving clip length.
- Watch minimums and expiry. Runway’s 28–80-credit per-generation minimums on Seedance, Veo’s whole 4/6/8-second clips and Kling’s expiring prepaid packages all raise the effective rate.
- Check discounts and regions yourself. Alibaba shows a limited-time 30% discount on Wan 3.0 (not applied here) and lists some models about 14% lower in Global-scope regions (no Wan 2.6 Flash there); we use the Singapore / International list price.
More context: the calculator, compare tools at 30s and 60s, AI video API pricing (all 138 rates per 5s–60s), AI video generator pricing comparison (apps and APIs), cheapest AI video generator, cheapest for YouTube, TikTok and Shorts cost, how much 1 minute of AI video costs, Wan API pricing, Kling API pricing, LTX API pricing, Grok Imagine cost, LTX vs Wan cost, Veo 3 cost per video, Seedance 2.5 cost and the FAQ.
Sources
Alibaba Cloud Model Studio pricing (page last updated Oct 22, 2026; checked Oct 24, 2026 — the page is rendered by JavaScript, so it was not re-fetched today), Google Cloud Agent Platform (Vertex AI) pricing (verified Oct 20, 2026, re-checked Oct 22, 2026) and the Veo 3.1 model page (checked Oct 25, 2026: Veo 3.1 Lite Preview, text- and image-to-video, 720p/1080p; Veo 3.1 Fast output 720p/1080p; fixed quota / Provisioned Throughput), BytePlus ModelArk pricing (Oct 22–23, 2026), Kling API video pricing and kling.ai/dev/pricing (Oct 25, 2026), LTX API pricing and the ltx.io API pricing page (re-checked Oct 26, 2026, same figures), xAI API pricing, the xAI models page and the xAI video overview (Oct 27, 2026: Grok Imagine Video 1.5 / 1.5 Lite prices, 1–15s, native audio, Lite 1080p upscaled from 720p), MiniMax video pricing (Oct 19, 2026; re-checked Oct 22) and Runway API pricing (Oct 22, 2026; Grok Imagine Oct 27, 2026; wan3_prime and kling2.5_turbo_pro Oct 25, 2026). Every rate lives in data.js / pricing.json or on AI video API pricing with its source. If a figure is not on an official page, we do not guess it.
FAQ
What is the cheapest AI video API in 2026?
On official list prices (checked Oct 24–27, 2026), the cheapest per-second AI video API is Grok Imagine Video 1.5 Lite at 480p on xAI's own API or the Runway API: $0.02 per second with native audio, so $0.60 per 30-second try and $1.80 for a usable 30-second clip with 2 retries (text- or image-to-video, 1 to 15 seconds per clip; xAI prices checked Oct 27, 2026). Next is Alibaba Cloud Model Studio's Wan 2.6 Flash at 720p without audio: $0.025 per second ($0.75 per try, $2.25 usable 30s; image-to-video only, Singapore / International list price). Then, tied at $0.03 per second ($2.70 usable 30s), Veo 3.1 Lite at 720p without audio on Google Cloud Vertex AI, LTX-2.3 Fast at 720p on LTX's own API and Grok Imagine 1.5 Lite at 720p, both with audio included; then Seedance 2.0 Mini 480p on BytePlus ModelArk ($0.036, $3.24) and Wan 2.6 Flash at 1080p without audio ($0.0375, $3.38). Kling's own API starts at $0.042 per second (Kling 2.6 or 2.5 Turbo at 720p without audio, $3.78).
What is the cheapest AI video API with native audio?
Grok Imagine Video 1.5 Lite at 480p on xAI's own API or the Runway API: $0.02 per second with native audio, so $0.60 per 30-second try and $1.80 for a usable 30-second clip with 2 retries (text- or image-to-video, 1 to 15 seconds per clip; xAI prices checked Oct 27, 2026). At 720p the floor is $0.03 per second ($0.90 per 30-second try, $2.70 per usable 30-second clip), shared by Grok Imagine 1.5 Lite 720p and LTX-2.3 Fast 720p on LTX's own API, which generates audio by default (LTX does not price audio separately; checked Oct 26, 2026). Next are LTX-2.3 Pro 720p at $0.04 ($3.60), then a three-way tie at $0.05 ($4.50): Veo 3.1 Lite 720p with audio on Vertex AI, Wan 2.6 Flash 720p with audio on Alibaba Cloud (image-to-video only) and Wan 3.0 480p on the Runway API. At 1080p with audio the cheapest is LTX-2.3 Fast at $0.06 per second ($5.40 usable 30s), then Wan 2.6 Flash at $0.075 ($6.75) and Veo 3.1 Lite at $0.08 ($7.20). Audio costs extra almost everywhere else: Wan 2.6 Flash doubles, Veo 3.1 Lite goes from $0.03 to $0.05 and Kling 3.0 from $0.084 to $0.126 per second at 720p.
What is the cheapest text-to-video API if I have no start image?
Grok Imagine Video 1.5 Lite at 480p on xAI's own API or the Runway API: $0.02 per second with native audio, so $0.60 per 30-second try and $1.80 for a usable 30-second clip with 2 retries (text- or image-to-video, 1 to 15 seconds per clip; xAI prices checked Oct 27, 2026); xAI's docs list text-to-video and image-to-video for both Grok Imagine 1.5 models. At 720p it is a tie at $0.03 per second ($0.90 per 30-second try, $2.70 per usable 30-second clip): Grok Imagine 1.5 Lite, Veo 3.1 Lite at 720p without audio on Google Cloud Vertex AI, and LTX-2.3 Fast at 720p on LTX's own API, which takes a text prompt or an image at the same price and includes generated audio (6 to 20-second clips, 16:9 or 9:16). Google's Veo 3.1 model page (checked Oct 25, 2026) lists text-to-video and image-to-video for Veo 3.1 Lite, which is still a Preview model, in 4, 6 or 8-second clips. Wan 2.6 Flash at $0.025 per second, second-cheapest overall, needs a first-frame image. Next among rates whose text-to-video support ClipCost has checked is Kling 2.6 or 2.5 Turbo at 720p on Kling's own API ($0.042 per second, $3.78 usable 30s); Seedance 2.0 Mini 480p on BytePlus is $0.036 per second, but check its input modes on BytePlus first.
What is the cheapest 1080p or 4K AI video API?
At 1080p with audio included, LTX-2.3 Fast on LTX's own API is $0.06 per second ($5.40 per usable 30s, text- or image-to-video). At 1080p without audio: Wan 2.6 Flash on Alibaba Cloud at $0.0375 per second ($3.38 per usable 30s, image-to-video only), then Veo 3.1 Lite on Vertex AI at $0.05 ($4.50) for text-to-video. Other 1080p rates with audio: Wan 2.6 Flash $0.075 ($6.75), then Veo 3.1 Lite $0.08 ($7.20). At 4K, the cheapest is LTX-2.3 Fast at $0.24 per second with audio ($21.60 usable 30s; 6, 8 or 10-second clips). Google's Vertex AI price list has Veo 3.1 Fast at $0.25 per second without audio and $0.30 with audio ($22.50 / $27.00 usable 30s), but Google's Veo 3.1 model page lists only 720p and 1080p output for Veo 3.1 Fast, so confirm 4K before you rely on it. Veo 3.1 Standard 4K is $0.40 without audio and $0.60 with it, and Kling 3.0 4K on Kling's API is $0.42 per second with or without audio.
Is the Runway API cheaper than going direct to the model maker?
Rarely. Runway's API charges $0.01 per credit, so Grok Imagine 1.5 Lite costs the same as on xAI's API ($0.02 / $0.03 / $0.14 per second at 480p / 720p / 1080p) while Grok Imagine 1.5 is cheaper on xAI ($0.08 / $0.14 / $0.25 against $0.10 / $0.16 / $0.29), and Wan 3.0 and Wan 3.0 Prime cost the same as on Alibaba Cloud ($0.05 / $0.10 / $0.20 and $0.068 / $0.14 / $0.28 per second at 480p / 720p / 1080p), and Veo 3.1 Standard costs the same as on Vertex AI ($0.40 with audio, $0.20 without), as does Veo 3.1 Fast without audio at 1080p ($0.10). Everything else is cheaper direct: Seedance 2.0 Mini is $0.07 / $0.15 per second (480p / 720p) on the Runway API against $0.036 / $0.076 on BytePlus, even after Runway's Oct 2026 API price cut, and Seedance 2.0 1080p is $0.40 against $0.374, Seedance 2.5 480p $0.20 against $0.103, HappyHorse 1.0 1080p $0.30 against $0.24 on Alibaba Cloud, Veo 3.1 Fast with audio $0.15 against $0.10–$0.12 on Vertex AI, and Kling O3 $0.13 / $0.17 / $0.43 against Kling 3.0 Omni with audio on Kling's own API at $0.112 / $0.14 / $0.42. The Runway API wins on convenience (one key, many models), not price.
Is the cheapest AI video API cheaper than an app subscription?
Yes, at the very bottom. Of the 210 verified rates ClipCost ranks per usable 30-second clip, seven of the nine cheapest are pay-per-second APIs: Grok Imagine 1.5 Lite 480p with audio ($1.80), Wan 2.6 Flash 720p and MiniMax H3-Lite 480P ($2.25), and Veo 3.1 Lite 720p without audio, Luma Ray3.2 540p on the Luma API, LTX-2.3 Fast 720p and Grok Imagine 1.5 Lite 720p with audio ($2.70 each). The cheapest subscription rate is Luma Ray3.2 540p on Luma Plus annual credits at $2.25 (tied for second since Nov 2, 2026; Kling 2.6 Standard mode follows at $2.70), and only if you use every credit before it expires at the end of the month. An API also costs nothing in a month you do not generate, but you need code (or a no-code wrapper), a cloud account and, for Veo, quota in a Google Cloud project.