ElevenLabs’ Creative Suite Stays in Beta as Flows Links 50+ Models
ElevenLabs’ visual creative suite remains available and officially in beta. Its current Image & Video documentation lists iterative image and video generation, audio-driven lip-sync, upscaling by as much as four times, Studio import, restrictions affecting some models and uploads in the United States, three daily image requests on the free plan and paid-only video generation.
The significant change since the original release is the addition of reusable, connected production pipelines. ElevenCreative Flows now links visual generation with ElevenLabs’ voice, music and sound tools, while the original proposition—creating and assembling several media types within one platform—remains intact.
The original launch consolidated tools rather than inventing every model
ElevenLabs introduced Image & Video as a beta on November 17, 2025. The company’s original Image & Video release identified Veo, Sora, Kling, Wan and Seedance as video options, alongside Nanobanana, Flux Kontext, GPT Image and Seedream for still images.
ElevenLabs did not build all those underlying visual systems. Its product placed third-party image and video models in the same environment as the company’s voice, music and sound-effect tools, giving creators a shorter route from a prompt or reference image to an assembled audiovisual project.
That distinction defines what “all in one” means here. It describes workflow consolidation, not one foundation model that produces every medium or a guarantee that all included models behave alike.
What the Image & Video workspace actually does
The workspace supports generation from text and visual references, followed by additional prompts and variations. Available inputs depend on the chosen model: a job may begin with text, a source image, start and end frames, or other reference material.
Enhancement and generation are separate stages. Media can be upscaled after creation, but an enhanced export should not be confused with footage generated natively at the final resolution; the source model’s own output settings still determine what enters the enhancement step.
Lip-sync similarly operates on an existing video and supplied audio rather than turning every visual model into an audio-video generator. This modular structure lets creators select a visual model, add a voice separately and then align the two, but it also leaves room for quality differences at each stage.
Completed assets can be downloaded or transferred to ElevenCreative Studio. Studio supplies an assembly layer for combining video, narration, captions, music and sound effects on a timeline, although specialist editing software may still be necessary for detailed compositing, color work or other advanced post-production.
Flows changes the suite from a workspace into a production system
Flows is the more consequential expansion because it preserves relationships between production stages. A node-based canvas can pass an image into animation, route audio into lip-sync and add speech, music or sound effects without requiring the creator to reconstruct the sequence for every new asset.
A June 2026 Flows Agent update describes a canvas connecting more than 50 image and video models with ElevenLabs’ audio models, plus a conversational agent that can choose models, create nodes and execute the resulting pipeline. The agent can also modify an existing chain when a creator requests a different voice, script, language or visual model.
This makes the platform more useful for recurring formats than the launch version was. A saved Flow can retain the structure of a production job while selected inputs change, so a creator does not have to document and repeat every handoff manually.
The agent does not remove the cost of generation. Its assist mode can pause before expensive operations and request approval, which matters when one instruction could otherwise trigger several credit-consuming visual stages.
The unified workflow still has firm boundaries
Beta status is the first boundary. The model roster, accepted inputs and availability rules can change, making a saved workflow more durable than an assumption that any particular model will remain accessible under the same conditions.
Regional access is another material limitation. A model shown in the broader product catalogue may be unavailable to a user in the United States, and restrictions can also affect uploaded reference material rather than generation alone.
Plan limits divide the image and video experience as well. Free access is suitable for a small number of still-image requests, but it does not provide video generation; the headline description of one multimodal suite therefore does not imply that every medium is included at every subscription level.
Model choice also remains a production decision rather than a cosmetic preference. Different systems accept different reference types and expose different duration, resolution, aspect-ratio, sound and credit options, so replacing one node may change both the output and the economics of a Flow.
What creators gain—and what they do not
The clearest gain is continuity between tasks that commonly require separate services. Visual references can remain connected to later stages, audio tools operate in the same ecosystem, and a successful sequence can be retained as a reusable pipeline instead of surviving only as an undocumented series of exports and uploads.
The trade-off is reliance on an aggregation layer over models with distinct technical and regional constraints. ElevenLabs reduces interface fragmentation, but it cannot make third-party systems interchangeable or guarantee identical availability, pricing and output characteristics.
The suite is therefore best understood as an orchestration and assembly environment. It has grown substantially beyond the initial image-and-video workspace, yet its current value comes from connecting production stages—not from eliminating model selection, credit management or specialist editing altogether.
Also read:
Subscribe to our newsletter
Get the latest Web3, AI, and crypto news delivered straight to your inbox.