ElevenLabs is transforming from a specialized voice AI provider into a comprehensive multimodal production hub, integrating high-end video models from OpenAI, Google, and Kling into its Studio platform.
Announced today, the update allows creators to generate visuals using restricted enterprise-grade models, including OpenAI’s Sora 2 Pro and Google’s Veo 3.1, directly alongside ElevenLabs’ native audio tools.
In a statement, the ElevenLabs Team emphasized that the update “unifies the most advanced AI models with our industry-leading voice, sound, and music tools,” effectively consolidating the best-in-class generative capabilities into a single subscription.
By aggregating third-party video generators within a single timeline editor, the company is positioning its Studio as a unified “Adobe for AI,” challenging fragmented workflows that force users to juggle separate apps for script, voice, and video production.
Aggregating the Giants: A New Multimodal Strategy
ElevenLabs has officially expanded its “Studio” platform to include Image & Video generation, marking a decisive shift from its roots as a pure-play audio AI company.
Rather than attempting to build proprietary video models from scratch to compete directly with incumbents like Runway or Luma, the company has adopted an aggregator strategy. This approach positions ElevenLabs as a unified interface layer for third-party giants, streamlining access to fragmented tools.
Included in the integration are some of the industry’s most coveted and restricted models. Users can now access OpenAI’s Sora 2 Pro and Google’s Veo 3.1, models that have seen limited public deployment outside of select partner programs.
This move positions ElevenLabs Studio as a direct competitor to traditional Non-Linear Editors (NLEs) like Adobe Premiere, but with a generative-first workflow that combines script, voice, sound effects, and visuals in one timeline.
By centralizing these tools, the company addresses the friction of the current AI creative stack. Typically, creators must generate assets across Discord, various web apps, and local software before assembling them elsewhere.
The platform now supports a seamless transition from text prompting to final video export within a single environment.
The pivot aligns with CEO Mati Staniszewski’s stated vision of building a “generational company,” moving beyond the commoditization risks of standalone text-to-speech services.
The expansion builds on a year of rapid growth and product diversification for the company. As previously covered by Winbuzzer, ElevenLabs recently doubled its valuation to $6.6 billion following a $100 million employee tender offer, signaling strong investor confidence in its broader platform strategy.
Under the Hood: Sora, Veo, and Kling Integration
For creators, the primary appeal lies in the specific capabilities of the integrated models. OpenAI Sora 2 Pro is positioned as the flagship video model, offering high-fidelity output at 720p or 1080p resolutions.
It supports fixed durations of 4, 8, or 12 seconds and is optimized for cinematic results and complex motion. However, this performance comes with a steep cost of 12,000 credits per generation and currently lacks support for end-frame references.
Google Veo 3.1 offers a professional-grade alternative focusing on creative control. It provides features such as negative prompts and dedicated sound control for 4-8 second clips at a cost of 8,000 credits. This model is particularly suited for users needing precise direction over the visual output rather than just raw fidelity.
Kling 2.5 is included for its strength in physics simulation and fluid dynamics. It generates 1080p video in 5 or 10-second bursts for 3,500 credits. While it lacks the sound control of the Google models, its lower credit cost and specific physics capabilities make it a viable option for dynamic scenes.
Beyond video, the platform integrates a diverse range of image models. Flux 1 Kontext Pro offers advanced style control, while Google Nano Banana is optimized for speed. Seedream 4 is available for generating consistent multi-shot sequences, crucial for storytelling projects.
To ensure these assets are viable for high-resolution displays, users can upscale outputs up to 4x using Topaz Upscale models. The suite also includes specialized utility models like Omnihuman 1.5 for animating static images and Veed LipSync for dubbing existing video.
These tools bridge the gap between visual generation and ElevenLabs’ core audio tech, which includes the recently launched Scribe v2 Realtime speech-to-text model.
Studio Workflow and the Cost of Creation
The Studio interface introduces a unified timeline where users can upload a video to auto-generate a script or write a script to generate corresponding visuals.
A key feature is the “Speech Correction” workflow, where editing the text transcript automatically regenerates the corresponding voiceover segment. This capability significantly streamlines the revision process, removing the need to re-record or manually splice audio.
Credit consumption is highly variable, creating a complex economy for users. A single high-end video generation using Sora 2 Pro costs 12,000 credits, significantly more than standard audio or image tasks.
The product guide notes that “video generation is only available on paid plans,” meaning free users are limited to image generation capabilities.
Export options are robust, supporting MP4 downloads with H.264/H.265 codecs and PNG for images. Alternatively, assets can be directly re-imported into Studio projects for further editing. The platform also supports “Image-to-Video” workflows, allowing users to use generated images as start frames to maintain visual consistency across video clips.
This aggregation strategy allows ElevenLabs to offer “best-in-class” performance across all modalities without bearing the massive R&D costs of training foundation video models itself.
It complements their existing portfolio, which includes the Eleven Music generation tool and the Voice Isolator API, creating a comprehensive ecosystem for AI-driven media production.
