Loading page…
Reviewed 2026-09-21. Every figure is either a public page we fetched that day, or a generation we ran ourselves and read back with ffprobe. We say plainly which job we do and which we don't — and the one we don't do is the one most of these tools are built for.
Search "make YouTube Shorts automatically" and you get one list. But the tools on it are doing two unrelated jobs, and picking from the wrong half wastes your month:
| Job | What you start with | Typical tools |
|---|---|---|
| Clip | A long video you already made | Opus Clip, Submagic, Flowshorts |
| Generate | A prompt or some images — no footage yet | Text/image-to-video models |
We do the second one. We do not do the first, and nothing below will pretend otherwise. If your job is "I have a 40-minute podcast, cut me ten Shorts," skip to the last section — the honest answer there is that we're not a candidate and we haven't tested the tools that are.
Fetched 2026-09-21, the way a first-time visitor and a crawler would: no login, no JavaScript.
| Tool | /pricing | Price in the served HTML | Machine-readable |
|---|---|---|---|
| Submagic | 200 | yes | ✅ $12 / $23 / $41 |
| Opus Clip | 200 | yes ($15 / $29) | ❌ |
| InVideo | 200 | yes ($12 / $101 / $122 / $150) | ❌ |
| Korpi | 200 | yes ($11 / $18 / $22) | ❌ |
| Flowshorts | 200 | no price string in the served HTML | ❌ |
| AIVG (us) | 200 | yes | ✅ six tiers, priceCurrency: USD |
What that last column does and doesn't tell you. If a price isn't in the HTML an automated reader receives, then whatever an AI assistant quotes you for that tool did not come from the tool's own page — go check before you budget on it. Flowshorts may well show a price to a human with a browser; what we can say is that a client without JavaScript doesn't get one.
We are not claiming machine-readable pricing means a better product. We looked for that link and the evidence ran the other way: tools with no structured pricing get recommended by assistants constantly.
Shorts work is volume work, so render time decides whether you iterate or wait. These are p50/p90 times from completed generations on our platform over a 30-day window. A model is quoted only with ≥5 samples and a p90/p50 spread under 2×.
| Model | Samples | Typical (p50) | Slow case (p90) |
|---|---|---|---|
| MiniMax H3 Max (camera) | 6 | 7s | 12s |
| Kling 3.0 Pro (image) | 8 | 128s | 250s |
| Seedance 2 Fast (reference) | 12 | 222s | 286s |
🔴 These are model-level numbers across all our traffic — not Shorts-specific jobs. Render time is driven by model, resolution and duration more than by what the clip is for, so they transfer reasonably, but they are not measurements of Shorts production and we're not dressing them up as such.
gemini-omni-1-1-text had 3 samples and a 2.6× spread, so we don't quote a
median for it. That exclusion is the rule working: a median is only useful when
the distribution behind it is tight.
Practical read: at ~7s, MiniMax H3 Max is the only one here you can iterate on in a single sitting — which is what a Shorts cadence actually demands. At 2–4 minutes the others are batch tools: queue a week of clips, come back.
Read back with ffprobe from the delivered file — never the request we sent, never what the provider's API reported.
| Model | Operation | Asked for | ffprobe read back | Match | Runs |
|---|---|---|---|---|---|
| Kling 3.0 Pro | image to video | 1080p / 5s | 1920×1080 / 5.042s | ✅ | 4 |
| Pikaframes | keyframe transition | 1080p / 10s | 1920×1080 / 10.042s | ✅ | 1 |
| Pikaframes | keyframe transition | 720p / 7s | 1280×720 / 7.083s | ✅ | 1 |
🔴 Runs column first. Two of three rows are a single observation, and the four-run row was measured on e-commerce clips, not Shorts. One run says a thing happened once, not how often. We publish them because a measured single run beats an unmeasured claim — not because they add up to a pattern.
Durations a few frames long are not a bug: encoders emit whole frames, so 7.083s at 24fps is 170 frames — 7s rounded up to a frame boundary.
Vertical note we have not measured: Shorts are 9:16. Every row above was measured at 16:9. We have no ffprobe data on whether these models honour a 9:16 request, and we're not going to guess from the 16:9 rows. If you're buying for Shorts, run one job and check the delivered dimensions yourself — it takes thirty seconds and it's the exact gap in our own data.
Then we are not a candidate, and we have not tested the tools that are.
Opus Clip, Submagic and Flowshorts are built for this: you hand them long-form footage and they pick moments, reframe to vertical and caption. We generate video from prompts and images; we have no clipping feature, so we have nothing measured to report about that job. The pricing column in section 1 is the only thing above that applies to them, and it's about their pages, not their output.
A list that ranks clipping tools it hasn't run is telling you about its author. We'd rather leave the gap visible.
The models in section 2 run on AIVG:
generate, download, ffprobe. If your reading disagrees with ours, ours is the
one that's wrong — tell us and we'll re-measure. Prices are on the
pricing page in real dollars.
Sources: pricing pages fetched 2026-09-21 without login or JavaScript;
measured-delivery-v1 (reviewed 2026-09-19); delivery_time_stats_v2
(generated 2026-09-20). The last two come from production generations, not
vendor documentation.