Feb 19, 2026
State of Generative Media Volume 1
At Prologue AI we use many of the models featured in this report every day. The State of Generative Media Volume 1 (2026) offers a comprehensive look at how generative technology accelerated in 2025. Here's our curated summary.
Introduction
From e-commerce teams to visual designers, anyone can now generate hundreds of production-ready images in minutes. A few years ago, this volume would have required thousands of photographers, studios, and production staff. The cost structure that has governed e-commerce and other digital verticals has shifted. Traditional barriers around content production are evaporating, due to generative media infrastructure.
The primary impact of breakthroughs in generative technology is the expansion of creative potential for users and builders alike. Entertainment applications initiated adoption of generative media, but in 2025 production applications (e.g. e-commerce, advertising, creative studios) drove scale and by year's end 88% of organizations deployed AI in at least one business function.
Jeffrey Katzenberg articulated the fundamental transformation:
It's the democratization of storytelling at a level that has never happened in the existence of humankind.
This shift emerged from rapid advances in generative technology, as models reached levels of quality, controllability, and reliability once reserved for specialized production teams.
This report examines how generative technology and trends accelerated in 2025. These insights draw heavily from survey data collected across a diverse range of organizations and individual users. Generative media is changing how we tell stories, build businesses and engage with users.
This signals the start of a new chapter in the digital age.
Model Maturity
In 2025, video generation models delivered outputs passing visual Turing tests for untrained observers. Technical capabilities advanced greatly across image, video, and audio generation in 2025, with different modalities reaching similar stages of evolution. Image editing capabilities revitalized a category that appeared to be declining. For all industries and modalities, infrastructure optimization reduced latency sufficiently for more real-time applications.
While individual launches defined key inflection points, the broader story of 2025 was scale. Model releases were no longer isolated breakthroughs. They became continuous across modalities:
| New models by modality* | Total endpoints |
|---|---|
| Video | 450 |
| Image | 406 |
| Audio | 59 |
| 3D | 35 |
| Speech | 35 |
| Total | 985 |
*New models across major platforms in 2025
Image generation performance
Image generation transformed experimental workflows into production pipelines in 2025. Black Forest Labs released Flux.1 Dev with superior prompt adherence, text rendering accuracy, and human pose fidelity. OpenAI's GPT Image 1 created a cultural moment with its Studio Ghibli aesthetic capturing billions of views across social platforms.
| Timeline | Model | Company | Impact |
|---|---|---|---|
| Aug 2024 | Flux.1 Dev | Black Forest Labs | Shattered performance ceiling, superior prompt adherence |
| Mar 2025 | GPT Image 1 | OpenAI | True multimodal image generation, defining cultural moment |
| Aug 2025 | Qwen Image Edit | Alibaba | Open-source image editing with LoRA |
| Aug 2025 | Nano Banana v1 | Google DeepMind | Consumer accessibility without technical proficiency |
Black Forest Labs launched Flux Kontext, the first dedicated image editing model achieving character consistency, style transfer, and localized editing at near-real-time speeds. ByteDance's Seedream 4.0 offered faster generation at lower computational costs while maintaining comparable output quality.
Video generation capabilities
Eight major video generation releases in ten months produced rapid competitive iteration. Google DeepMind released Veo 2 in December 2024, establishing physically accurate video as the quality benchmark. The model's physics simulation accurately modeled gravity, water dynamics, and object interactions.
| Timeline | Model | Company | Key Innovation |
|---|---|---|---|
| Dec 2024 | Veo 2 | Google DeepMind | Physically accurate video, quality benchmark |
| Apr 2025 | Kling 2.0 | Kuaishou | First-frame-last-frame narrative control |
| May 2025 | Veo 3 | Google DeepMind | Native audio generation with video output |
| Jul 2025 | MirageLSD | Decart | Live-stream diffusion, real-time generation |
| Sep 2025 | Sora 2 | OpenAI | Scene-aware multi-modal generation |
Kling 2.0 in April 2025 introduced first-frame-last-frame functionality, giving creators precise narrative control over generated sequences. Veo 3 enabled rapid-turnaround workflows for social media and content channels. Sora 2 combined native audio with excellent multi-shot generation in a single output.
Major releases arrived every 4–6 weeks in 2025.
Voice, music and audio synthesis
Audio became one of the most production-ready generative media categories in 2025. ElevenLabs Turbo v2.5 is among the most widely used low-latency text-to-speech systems (~250–300ms), while MiniMax Speech-02 achieved 99% human voice similarity across 32 languages. As one generative voice user noted: "Sub-300ms is table stakes for voice AI. Above that, the experience breaks."
Eleven Music by ElevenLabs (August 2025) was the first major AI music model trained entirely on licensed data, establishing opt-in participation and 50/50 royalty splits for artists.
3D modeling and world models
3D generation matured from experimental outputs to production assets in 2025, compressing modeling timelines from weeks to minutes. Tencent released Hunyuan 3D 2.0, Meshy launched v5 and v6 preview, and Microsoft's TRELLIS 2 can generate high-resolution assets in under 3 seconds.
World models simultaneously generate and simulate interactive 3D environments where all modalities converge. DeepMind announced Genie 2, generating playable 3D environments from single image prompts. Fei-Fei Li's World Labs launched Marble in November 2025, the first commercial world model product.
State of Adoption
Enterprise generative AI adoption accelerated through 2025. Personal users bypassed technical requirements through emerging consumer applications. Organizations faced distinct barriers: model orchestration complexity, integration decisions and cost management. Businesses used two pathways for access: applications (65%) and APIs (62%), with many using both.
44% of image generation is in production workflows, compared to 39% for video.
| Industry Vertical | Adoption Rate | Primary Use Cases |
|---|---|---|
| Advertising | 56% | Rapid creation of campaign visuals, banner ads, social graphics at scale |
| Entertainment, Media & Creative Storytelling | 43% | Storyboarding, pre-visualization, special effects, short-form promotional clips |
| Creative Software or Tools | 31% | Design platform, creative software, video/image editing tools |
| Educational and Training Content | 30% | Interactive learning videos, animated explainers |
| Retail & E-Commerce | 19% | Automated product photography, catalog images, virtual try-on mockups |
| Architecture & Real Estate | 8% | 3D renders, staging visuals, and concept imagery for developments |
Source: Artificial Analysis (2025). State of Generative Media Survey Report 2025
Enterprise ROI
Return on generative media investment materialized faster than expected:
| ROI Status | Percentage of Organizations |
|---|---|
| Already profitable | 34% |
| Expecting returns within 12 months | 31% |
| Total achieving ROI ≤ 12 months | 65% |
74% of companies report their initiatives meet or exceed ROI expectations. Organizations achieving generative scale made structural changes: 43% redesigned workflows and production pipelines, 33% invested in staff training, and 30% allocated dedicated budget for media generation infrastructure.
Industry snapshots
-
Advertising: 75% adoption, up from 61% in 2024. Legal concerns (IP ownership, liability) dominated hesitation. Agencies achieving scale used generative media for content variation and A/B testing rather than primary asset creation.
-
E-commerce: Product image generation became a core infrastructure capability. Matt Koenig (Shopify): "The creativity of models absolutely cannot interfere with product fidelity."
-
Film & TV: Major studios allocated less than 3% of production budgets to generative AI. Media companies' AI spending projected to grow at 37.2% CAGR from $2.6B to $12.5B (2024–2029).
-
Gaming: 68% actively implementing AI. 40% of studios experienced productivity gains exceeding 20%. Burkay Gur: "Text-to-game will be the continuation of text-to-video; it's essentially making the video output interactive."
-
Education: One of the largest untapped opportunities. Sonya Huang (Sequoia): "The challenge is the bottleneck to create high quality content at scale that is most ideal for the learner."
Developer Experience
Infrastructure quality became a determining factor in development velocity during 2025. Organizations successful in scaling prioritized optimized serving infrastructure over model selection.
| Decision Criterion | Organizations Prioritizing |
|---|---|
| Cost optimization | 58% |
| Model availability | 49% |
| Generation speed | 41% |
| Reliability and uptime | 37% |
| Data security and compliance | 34% |
Enterprise production deployments use a median of 14 different models. The belief that single "omni models" would handle all generative tasks proved incorrect — task-specific optimization consistently outperformed general-purpose approaches.
Future Generations
The trajectory of generative media development through 2026+ is clear. Three major themes will dominate:
- Multimodal advancements (e.g. world models)
- Infrastructure optimization
- Democratization of creative tools
Expertise will become orchestration rather than execution. Taste becomes scarce, while capability becomes abundant. As technical capabilities are more commoditized, the fundamental value proposition shifts. "It's the storytelling that matters."
Solo entrepreneurs will generate visual content indistinguishable from large-scale production companies. The durable competitive moats will belong to teams that understand how to best deploy generative media, now that generating professional media is easier than ever.
State of Generative Media Volume 1, published 2026. In partnership with Artificial Analysis.
Prologue AI Team