Runway Aleph

What Aleph is

Runway Aleph is a video model that Runway introduced on 25 July 2025. Runway describes it as a state-of-the-art in-context video model for multi-task visual generation.^[1] The term "in-context" refers to how the model is used rather than to a separate feature: the material supplied together with a request serves as the context that conditions the result. For a video model, that material is a video, and the output is a new video derived from it. The design point is that one model is asked to handle many different jobs rather than a single narrow effect.^[1]

What it can do

Runway groups the operations into a few kinds. The first is editing at the level of objects: Aleph can add objects to an input video, remove them, or transform them.^[1] Adding means putting something into the footage that was not there; removing means taking something out; transforming means altering an object that is already present. The second is viewpoint: the model can generate a new angle of a scene.^[1] Footage recorded from one position can therefore come back from a different one. The third is appearance: the model can change the style or the lighting of a scene.^[1] Style and lighting sit apart from what is in the frame, so an edit of this kind changes how the footage looks without necessarily changing what happens in it.

Beyond these examples, Runway says Aleph can perform a wide range of edits on an input video, among many other tasks.^[1] Each of the listed operations — a new camera angle, different lighting, an object that appears or disappears — corresponds to work that would otherwise be done on set or in post-production, which is why a single model covering several of them at once is the claim being made.

Who can use it

Aleph is available to all paid Runway users.^[1] It is a released product rather than a demonstration reserved for researchers: anyone holding a paid Runway account can use it.

Where it sits

The emphasis in Runway's description falls on the combination — generation and manipulation of video, many tasks, one model. Multi-task here means the same model is applied to several different operations, in contrast with a collection of separate tools each built for one effect. Runway calls Aleph a novel step forward for multi-task video generation and manipulation.^[1]

Aleph also reached beyond Runway's own product. On 16 December 2025, TechCrunch reported that users can give Adobe Firefly specific instructions by using Runway's Aleph model.^[2] In that arrangement the model works inside another company's creative tool: the user states what should change, and Aleph carries out the instruction on the video. The two dates mark two stages — the model's introduction in July 2025, and its use inside a widely known editing product by the end of that year.