The condition printed beside Sora's rate
Sora publishes a batch tier at half the standard figure and states that it applies on every model and resolution. That halves the whole grid, including the dearest unconditional second in this register. As of 2026-09-22.
| What is published | As the API price page states it |
|---|---|
| The condition | The batch tier is half the standard figure |
| What it applies to | Every Sora model and every resolution on the page |
| Standard grid underneath | Ten cents a second up to seventy by model and resolution |
| The grid under the condition | Five cents a second up to thirty-five |
Inclusion rule. Conditions the API price page applies to its own figures, with the scope the page gives them. A condition attached to one model would be recorded at that scope instead. Order. The condition first, then its scope, then the figures.
1Where Sora sits in this column
| Channel | Condition on the figure |
|---|---|
| Runware | A struck-through cell on one model |
| Sora | Half price on the batch tier |
| Vidu | Half price off peak |
| WaveSpeed AI | Every figure may move with settings |
Inclusion rule. Channels that print a condition on the same page as the rate: a discount, a tier or a statement that the figure varies. A bare figure with nothing attached is not listed. Order. Alphabetical by channel name.
2A halving with universal scope is rarer than a halving
Most discounts in this register attach to one tier, one plan or one model. This one is stated across the whole grid, which means it does not change the internal ranking of the page at all: every row moves by the same factor.
That is what makes it usable in planning. A production can decide separately which model and resolution it needs and whether the work can wait, and the two decisions do not interact.
3What a buyer trades for the other half of the money
A batch tier trades latency for price. The page prices it and does not promise a turnaround, so the cost is an unbounded wait rather than a stated one. For interactive work that is unacceptable; for overnight generation of a season's segments it is close to free.
Read that way, the standard tier is the price of a same-day answer and the batch tier is the price of the video. Most of a serial pipeline only needs the second.
4Under the condition, the dearest row stops being extraordinary
Seventy cents a second unconditional is the top of the whole ledger. Thirty-five is expensive and ordinary, sitting near the standard tier of a couple of other channels. The condition therefore changes how this channel should be described, not just what it costs.
Which is why conditions get their own column here. A rate quoted without the string attached to it can be out by a factor of two, and the string is on the same page as the number in every entry in this column.
- The batch tierHalf the standard figureon every Sora model and resolution
- sora-2 at 720p$0.10 per second
- sora-2-pro at 1080p$0.70 per secondthe dearest published second in this ledger
5Inside a generated second
Price aside, what goes into the seconds behind “Sora price condition” is described in SceneMixer's guides:
Measured across every job that finished successfully on scenemixer.com in the thirty days to 20 September 2026: analysis of a whole book runs about 2 minutes 49 seconds at the median, one episode's storyboard 1 minute 54 seconds, one video segment 3 minutes 20 seconds, and compositing an episode 3 minutes 52 seconds.
SceneMixer, Turn a Novel into a Video Series with AI — “How long does it take to produce one episode?” (read 2026-10-13)
Each character is matched to a preset voice sample that is attached to every segment, so the character's voice identity is stable across episodes.
SceneMixer, Native-Language Dialogue vs. Dubbing in AI Video (2026): Why the Cast Should Speak the Line — “Does the character keep the same voice across languages?” (read 2026-10-13)
6Sources
Read from the API price page published at developers.openai.com on 2026-09-22. The rate this column sits beside is on Sora; what the register accepts as published is on the method page. The same column on other channels: Runware, WaveSpeed AI. Every column: what each channel publishes.