As reviewed on 28 October 2026, ChatCut’s omni selector maps to Gemini Omni 1.1 Flash, public ID gemini-omni-1.1-flash-preview. It is separate from Veo and Kling Omni. The pinned contract offers 3–10-second clips at 360p/720p/1080p/4k; the saved guide labels the last two tiers upscaled. This reference describes exposed controls; it contains no new successful Omni generation or production-access proof.
Which model is this?
A short alias is not a model family. Keep the exact public ID with any saved prompt or quote, and do not transfer Veo’s durations, audio-reference rules or first/last-frame contract to Omni. The Google Veo model reference documents a separate family; the ChatCut model matrix explains that distinction.
| Model and public ID | Facts and evidence |
|---|---|
Gemini Omni 1.1 Flashgemini-omni-1.1-flash-previewAvailability: Pinned integration and saved dev checks; production unverified. Source links describe native specifications. Reviewed: | Native documentation: Gemini Omni 1.1 Flash, distinct from Veo or Kling. Public documentation and pinned contract identify this exact preview model; no new native boundary test. ChatCut integration: omni alias; 3–10s; 9:16/16:9; 360p/720p/1080p/4k. Saved guide labels 1080p/4K upscaled. First(+last) frames, up to 3 image references, or short-source edit/extend via continueFrom. Endpoint evidence: Pinned contract and guide only; no new successful Omni generation or production entitlement check. |
The source for the controls above is the pinned ChatCut product contract and saved video-tool guide, reviewed 28 October. The public Google Omni reference, checked 28 October, confirms the exact identity and documented control families. ChatCut controls remain scoped to the pinned contract, and no new native boundary or endpoint output was tested. Source-code exposure, account entitlement, successful output and current deployed availability are separate checks.
What inputs and settings are exposed?
The inspected contract lists 3–10 seconds, fixed 16:9 or 9:16, 720p default, and 360p/720p/1080p/4k output labels. The saved guide identifies 1080p and 4K as upscaled; those choices do not prove native 4K detail. Exporting a timeline at 4K is also a separate operation from generating or upscaling the source clip.
Choose the applicable path: a first frame, optionally a last frame; up to 3 reference images; or a short source video through continueFrom for an edit or extension. Treat the source slot and image-reference mode as different paths, rather than assuming every input can combine. The saved guide describes source clips up to 10 seconds; the pinned contract allows 10.6 seconds of measured source duration as an encoding tolerance. This tolerance is not a documented 10.6-second output tier. Do not infer video/audio-reference arrays from the word “Omni.”
For public resource inputs, ChatCut’s inspected handler uses 5 MiB per image and 50 MiB per video/audio before applying model rules. Prefer an existing project asset when available, and verify the input slots after changing models. Generation references covers asset roles and incompatible-mode diagnosis.
Make a short clip, then request one change
Choose one small request
Choose Gemini Omni actually shown in Video Gen. Set a five-second 16:9 request at 360p or 720p and use a simple subject and camera action.
Assign the reference
For an image-led clip, attach one clear image and say: “Use this image for the product silhouette and color; show it turning slowly on a plain gray surface; locked camera; no text.” These are requested constraints, not guarantees.
Revise a short source
After inspecting a completed clip, select the exposed edit path or ask the Agent to use that clip as continueFrom. Request one change, such as a background color, and explicitly keep the subject, motion and framing. Check every inherited setting and the new quote.
This is an unexecuted example, not a success claim. Completion should give you a new video asset to review and place on the timeline; it does not make generated objects separately editable. Multiple edits may change details you asked to preserve. No fixed safe number of edit rounds, quality ranking or speed guarantee was established by the supplied evidence.
Does a 360p draft become a matching final?
“Draft” here means an ordinary low-resolution generation. The inspected Omni control is not a Seedance Draft task ID or a promise of frame-identical rerendering. Before paying again for a higher tier, check whether you need another generation, a source-based edit or an upscale, and inspect the actual resulting dimensions. For native Seedance Draft-to-final and its current ChatCut exposure gap, see the video controls tutorial.
Credits and route failures
Use the current preflight quote and Credits Policy. Compare exact model ID, cohort/catalog, ratio, resolution, requested output duration, source duration and quote date. Reconcile the settled charge afterward; Agent usage is separate. The contract’s estimation formula is not a current customer invoice, and this page has no verified current credit example.
If Omni is absent or a job is rejected, check the visible model, source duration, input mode and workflow failures. A timeout or missing option does not prove the native model lacks the capability. Do not switch to Veo or Kling under the same alias.
Frequently asked questions
Is ChatCut’s Omni selector Veo 3.1?
Does the 4K option mean native 4K generation?
Can I use a 15-second source clip?
Can I attach reference audio because the model is called Omni?
Was this exact Omni route successfully tested?
Last verified: 28 October 2026. Evidence scope: pinned ChatCut Omni contract, saved video-tool guide and model mapping, reviewed on that date; no new generation. Public sources: Google Omni reference, checked 28 October, and the related Google Veo reference, accessed in the supplied research 27 October, supports only the separate Veo family. Continue with video generation, the model matrix or generation credits.