Gemini + Sora Native Video? We Checked. It’s Not Real.
Claims about Gemini 2.5 Pro generating videos via Sora’s SDK don’t survive basic fact-checking. Here’s what Google has actually shipped.
A story has been circulating that Google enabled native 60-second video generation in Gemini 2.5 Pro through a Sora SDK integration — complete with a suspiciously precise “40% faster iteration cycles” stat. Sounds exciting. The problem: none of it checks out. Not the model name, not the integration, not the number.
Promptyze ran this through official Google documentation, the Google Cloud Blog, and OpenAI’s published Sora resources. Here’s what the record actually shows.
Gemini 2.5 Pro Doesn’t Exist Yet
As of February 2026, Google has not officially released a model called Gemini 2.5 Pro. The current flagship lineup sits at Gemini 2.0 Flash — announced in December 2024 — along with Gemini 1.5 Pro and Flash variants still in wide use. Google has confirmed that Gemini 2.0 Flash brings native multimodal support covering text, image, and audio, but that’s a different product with different capabilities than what was described in the original claim.
There is no entry on the Google Cloud Blog, no Gemini release note, and no developer changelog referencing a Gemini 2.5 Pro launch. If it existed, Google would have published something — they tend not to stay quiet about model upgrades.
Sora and Gemini Are Competitors, Not Partners
Sora is OpenAI’s video generation tool. Gemini is Google’s AI platform. The idea that OpenAI would ship a public SDK for a deep integration with a direct competitor’s product — without either company announcing it — would be a genuinely strange business decision. OpenAI’s Sora API access remains limited and tightly controlled. There is no public documentation from OpenAI describing an enterprise partnership or SDK release targeting Gemini.
Google’s own video generation work runs through Veo 3, its in-house model. Veo 3 is where Google is actually investing in long-form, high-quality video output — not through a licensing deal with OpenAI’s tooling.
Where Did the 40% Figure Come From?
No source. That number appears in the original brief with zero attribution — no study, no Google benchmark, no independent test. Round percentage figures with no methodology attached are almost always a red flag, and this one delivered exactly that. It got left out of this article for the same reason it shouldn’t appear in any article: it doesn’t trace back to anything real.
What Google Is Actually Building
To be fair to the underlying idea, Google is genuinely pushing toward bundled multimodal capabilities. Gemini 2.0 Flash handles text, image, and audio natively in one model. Veo 3 handles video generation separately but within the Google ecosystem. NotebookLM continues expanding its audio and document features. The direction — consolidating these into a unified product experience — is real. The specific claim about Sora SDK integration and Gemini 2.5 Pro is not.
Google has also been testing video capabilities in AI Studio and experimenting with Workspace integrations, but those features go through Veo, not a third-party tool from a competitor.
Why This Matters
AI news moves fast enough that fabricated or unverified product claims spread before anyone checks them. A plausible-sounding integration between two well-known AI products, dressed up with a specific performance stat, is exactly the kind of thing that gets screenshot and reshared without a second look. Promptyze’s job is to be the second look. This claim failed it. If and when Google actually ships Gemini 2.5 Pro with native video generation — through Veo or anything else — it’ll be here with sources attached.
