Why Sora is unrated on cost
Sora is one of the two rows left unrated for cost transparency, which sits oddly beside the fact that this register carries three published per-second figures for it and a batch tier halving them. The two statements are about different pages. As of 2026-09-12.
| Point | As the rubric has it, or as the table records it |
|---|---|
| The question this column asks | Can the cost of a finished minute be worked out before starting? |
| What this tool rates | Not rated |
| Tools holding the same value | Two of the sixteen rated here |
| What a five looks like | A public per-second or per-credit rate with the credit price beside it |
| What leaves the column unrated | Price shown only after sign-in, or only as a monthly allowance |
| Tools left unrated on this column | Two of the sixteen |
Inclusion rule. The rubric's own wording for this criterion, with this tool's value beside it. Nothing here is taken from a demo, a benchmark or a review. Order. The question first, then the value, then the rule it was given under.
1The whole column, with this row marked
| Tool | Cost transparency |
|---|---|
| CapCut | 4 of 5 |
| Dreamina | 3 of 5 |
| Hailuo (MiniMax) | 4 of 5 |
| invideo AI | 3 of 5 |
| Kling AI | 3 of 5 |
| LTX Studio | Not rated |
| Luma Dream Machine | 2 of 5 |
| Morph Studio | 4 of 5 |
| Pika | 4 of 5 |
| PixVerse | 3 of 5 |
| Runway | 2 of 5 |
| SceneMixer | 5 of 5 |
| Sora (OpenAI) | Not rated |
| Veo / Flow (Google) | 2 of 5 |
| Vidu | 3 of 5 |
| Wan (Tongyi Wanxiang) | 4 of 5 |
Inclusion rule. Tools rated on this rubric: a public English product page addressing at least four of the seven criteria. An unrated cell means nothing usable is published on that point. Order. Alphabetical by tool name.
2The rubric reads the product page, the ledger reads the rate page
This column asks whether a production can work out the cost of a finished minute from what the tool itself publishes. For the consumer-facing product no usable rate appears, so the cell is blank. The developer price page for the same model family publishes a full grid, and that is where the ledger figures come from.
Both readings are correct and they answer different questions. A buyer using the app cannot read a rate; a buyer calling the API can read three of them plus a batch modifier that halves all of them.
3Why the blank is not scored low
Scoring an unpublished price as a two would put it beside tools whose rates are merely hard to read. The difference matters to a reader: one situation is solved by reading more carefully and the other is not solved at all from public material.
This row is also averaged over six criteria rather than seven for the same reason. Comparing that average against a fully rated one is the mistake the per-column pages are meant to prevent.
4The row carries a date that the ratings do not
The vendor has published end dates for this line, which no other entry in the table has done. None of the seven values changes because of that, and what it changes is how long any of them will be true.
A high picture-quality value on a product with a published ending is a short-term fact. Ratings describe what is published today; dates are what make them perishable, and this is the row where that is explicit.
Ratings describe what each vendor publishes on its own pages, read 2026-09-12, and are given under the rule on how scoring works. This column across every tool is on cost transparency; this tool across all seven is on Sora (OpenAI). Nearby cells: SceneMixer, script handling, LTX Studio, script handling. All of them: single ratings.