Sora Is Gone, Veo Is Free, and AI Video Is Now a Platform War
OpenAI shut down Sora in April, Google opened Veo 3.1 for free, and Kling AI hit the App Store top 5 — AI video has completely reshuffled.
On this page
Four months ago, OpenAI’s Sora was the name everyone associated with AI video generation — the model that made filmmakers nervous and marketers excited. Today, the Sora app is gone, its website is dark, and the model has vanished from leaderboards entirely. In its place: a free Google product available to everyone, a Chinese app climbing the App Store charts, and a ByteDance project that generates audio and video together. The AI video category did not just shift — it reshuffled from scratch.
What happened to Sora?
The backstory matters for understanding why the field looks the way it does now. On April 26, 2026, OpenAI shut down the Sora app and sora.com. The Sora API is scheduled to follow on September 24, 2026, completing the wind-down. Reports indicated the core problem was economics: Sora was resource-hungry and free for users, a combination that made it financially unsustainable at the scale OpenAI was operating.
By August 2026, the disappearance is complete. Sora models no longer appear at all on Artificial Analysis’ crowd-voted text-to-video leaderboard — not as a low-ranked option, not as a deprecated entry. Gone. For a model that defined the category’s public imagination less than a year ago, the speed of the erasure is striking.
Sora’s shutdown is a case study in the risk of launching an expensive generative AI product without a sustainable pricing model. The model’s quality was never the issue — its computational cost was. Offering a resource-intensive video generation model for free created demand that accelerated losses rather than building toward revenue. The takeaway for every AI lab: capability without viable economics is a demo, not a product.
The April shutdown is not August news, and it is important to be clear about that timeline. But Sora’s absence is the context that makes everything happening now in AI video comprehensible — it left a vacuum that the rest of the market has spent four months filling.
Who is filling the vacuum?
Three players, each with a different strategy.
Google: Veo 3.1, free for everyone
Google’s response to Sora’s exit has been the most aggressive distribution play in AI video to date. Veo 3.1 is now available to everyone for free, pushed through an unusually broad set of surfaces: Gemini, Flow, Google Vids, Google AI Studio, the Gemini API, and Vertex AI. That is six different entry points spanning consumer (Gemini), creative professional (Flow, Vids), developer (AI Studio, API), and enterprise (Vertex AI) tiers — all at zero cost.
The strategy is unmistakable: where Sora tried to be a standalone destination and ran out of money, Google is embedding video generation into products people already use and absorbing the infrastructure cost across its cloud business. It is the classic Google playbook — make the capability free, distribute it everywhere, and let the value accrue to the ecosystem rather than charging for the feature itself.
Kling AI: the App Store contender
While Google is playing the distribution game, Kling AI from Kuaishou has been building a consumer product. By August 2026, Kling climbed to the number 5 spot on the App Store, a position that puts it alongside mainstream consumer apps rather than niche creative tools. Kling AI 3.0 is described as becoming a serious China-led video platform — not just a model you can call through an API, but a full product experience.
Kling’s rise is significant for two reasons. First, it demonstrates that AI video generation has crossed the threshold from “interesting technology demo” to “app people download and use regularly.” Second, it shows that the competitive field is genuinely global — the category is not going to be decided by US-based labs alone.
ByteDance: Seedance 2.0 and native audio-video
The most technically interesting move is coming from ByteDance. Seedance 2.0 is pushing toward native multimodal audio-video generation — generating synchronized audio and video together in a single pass rather than producing video and then adding or generating audio separately.
This matters because the current standard workflow for AI video is fundamentally disjointed: generate the visuals, then either find audio that fits or generate audio separately and hope it matches. The result is often subtly (or not so subtly) wrong — footsteps that don’t match walking speed, ambient sounds that feel detached from the scene, music that starts and stops at arbitrary points. Native audio-video generation aims to solve this by treating sound and image as a single output rather than two outputs stitched together.
Available through six surfaces (Gemini, Flow, Google Vids, AI Studio, API, Vertex AI) at zero cost. The bet: embed video generation into existing products and absorb the cost rather than building a standalone destination. Widest accessibility of any current AI video tool.
A consumer-first product from Kuaishou that has climbed to the App Store top 5. Demonstrates that AI video has moved past the demo-and-API phase into mainstream app territory. The leading China-led contender in the consumer AI video space.
How has the competitive landscape actually changed?
The shift is structural, not just a reshuffling of names on a leaderboard. When Sora launched, the AI video narrative was simple: one breakthrough model that was clearly ahead of everything else, made by the leading AI lab, generating astonishment and anxiety in equal measure. That narrative is gone.
What replaced it is a fragmented platform war spanning at least five distinct competitive arenas:
- Cinematic generation
High-quality, long-form video generation for filmmaking, advertising, and professional content. Veo 3.1 and Kling 3.0 are both competing here, with very different distribution strategies.
- Consumer/social video
Quick, accessible video generation for everyday users and social media creators. Kling’s App Store position puts it squarely in this space. Google’s Gemini integration does the same from a different angle.
- Enterprise creative suites
Video generation as a feature within broader creative and productivity tools. Google Vids and Vertex AI integration represent this approach — video as a capability within a workflow, not a standalone product.
- Multimodal generation
Audio-video generation as a unified output. Seedance 2.0 is the clearest play in this direction, aiming to eliminate the gap between separately generated audio and video.
- Developer APIs
Video generation as an API call for developers building applications. Google’s AI Studio and Gemini API, plus the remaining months of Sora’s API (until September 24), serve this market.
What does this mean for people building with AI video?
The practical upshot is that AI video generation is no longer a one-vendor story, and the “best” choice depends entirely on what you are building.
If you need the widest accessibility and zero upfront cost, Veo 3.1 through Google’s ecosystem is the obvious starting point — it is free, available through multiple interfaces, and backed by infrastructure that is not going to run into Sora’s sustainability problem.
If you are building a consumer product and care about what your users are actually downloading, Kling AI’s App Store traction is a signal worth watching — it tells you which video generation experience the public is gravitating toward right now.
If your use case requires synchronized audio and video — product demos, narrative content, anything where sound matters — Seedance 2.0’s multimodal direction is the most relevant technical development, even though it is earlier in its maturity curve.
| Platform | Strategy | Key Strength | Distribution |
|---|---|---|---|
| Veo 3.1 (Google) | Free, embed everywhere | Zero cost, six entry points | Gemini, Flow, Vids, AI Studio, API, Vertex AI |
| Kling AI 3.0 (Kuaishou) | Consumer app product | App Store #5, mainstream UX | iOS/Android app |
| Seedance 2.0 (ByteDance) | Native audio-video | Synchronized multimodal output | In development |
| Sora (OpenAI) | Shut down (Apr 2026) | — | API until Sept 24, 2026 |
Do
- Evaluate AI video tools by distribution model and pricing sustainability, not just output quality — Sora proved that quality without economics does not last
- Plan for a multi-vendor landscape: pick tools by use case (consumer, enterprise, developer, multimodal) rather than assuming one winner will dominate
- If you have active Sora API integrations, start migration planning now — the API shuts down September 24, 2026
Don't
- Don’t assume the current leaders are permanent — the category reshuffled completely in four months, and it can do so again
- Don’t ignore Kling and Seedance because they are from Chinese companies — the App Store rankings and technical innovations are real regardless of origin
- Don’t conflate “free” with “permanent” — Veo 3.1 is free today, but Google’s pricing decisions can change as the market matures
The AI video generation category in August 2026 looks nothing like it did in January. The model that defined the space is gone. The company that built it has moved on. And the replacements are not a single successor but a fragmented competitive landscape where the winners will be determined not by who generates the most impressive demo clip, but by who builds the most sustainable product around the technology. Sora proved the capability was real. What is happening now proves the business model matters more.
Frequently asked questions
Why did OpenAI shut down Sora?
OpenAI shut down the Sora app and sora.com on April 26, 2026, with the Sora API scheduled to end September 24, 2026. Reports indicated Sora was resource-hungry and offered free to users, making it financially unsustainable at scale. As of August 2026, Sora models no longer appear at all on Artificial Analysis' crowd-voted text-to-video leaderboard.
Is Google Veo 3.1 free to use?
Yes. Google opened Veo 3.1 to everyone for free, distributing it through Gemini, Flow, Google Vids, Google AI Studio, the Gemini API, and Vertex AI. This broad, zero-cost rollout across both consumer and developer surfaces makes Veo the most widely accessible high-quality AI video generator currently available to the general public.
What is Kling AI and why is it significant?
Kling AI is a video generation platform from Kuaishou, a major Chinese tech company. By August 2026 it climbed to the number five spot on the App Store, and its Kling AI 3.0 release is described as becoming a serious China-led video platform. Its rise demonstrates that AI video generation competition now spans both Western and Chinese markets.
What makes ByteDance's Seedance 2.0 different from other AI video tools?
Seedance 2.0 from ByteDance is moving toward native multimodal audio-video generation, meaning it can generate synchronized audio and video together in a single pass rather than producing them separately and stitching them afterward. This approach could eliminate the awkward mismatch between generated visuals and added audio that plagues current AI video workflows.
/* Comments */
Comments are offline right now — we reconnect automatically, nothing is lost.