Whether you're looking for an angel investor, a growth advisor, or just want to connect — I'm always open to great ideas.
Get in TouchOpenAI is preparing to integrate its Sora AI video generator directly into ChatGPT after the standalone app saw installs plummet 45% in January — a major pivot that could bring video generation to hundreds of millions of users.
AI, startups & growth insights. No spam.
TL;DR: OpenAI is planning to integrate its AI video generator Sora directly into ChatGPT after the standalone app suffered a 45% month-over-month decline in installs in January 2026, following an earlier 32% drop in December. The Information first reported the integration plans on March 11, 2026. OpenAI has not officially confirmed a timeline, but the strategic rationale is straightforward: ChatGPT's 900 million weekly active users represent an instant distribution channel that dwarfs anything a standalone video app could realistically capture. The Sora standalone app is expected to continue operating alongside the ChatGPT integration.
OpenAI launched Sora as a standalone iOS app in October 2025 to enormous fanfare — 100,000 downloads on day one, 1 million installs faster than ChatGPT itself. By January 2026, monthly installs had fallen to 1.2 million, down from a December peak and continuing a steep post-launch slide. A Disney partnership failed to arrest the decline. The numbers told OpenAI something the hype had obscured: video generation as a standalone product is a hard sell. As a feature inside the world's most-used AI product, it becomes something else entirely.
The 45% figure needs context to be understood properly. Sora was not a failing app before January — it was a wildly successful launch that experienced an entirely predictable post-viral correction.
When Sora launched on iOS in October 2025, it hit 100,000 downloads on its first day despite requiring invites. Appfigures data shows Sora accumulated 9.6 million total downloads across iOS and Android by the time the decline story broke, with $1.4 million in consumer spending. The U.S. accounted for $1.1 million of that figure, with Japan, Canada, South Korea, and Thailand rounding out the top markets.
The slide began almost immediately after peak. December 2025 saw installs fall 32% month-over-month. January 2026 added another 45% drop, bringing monthly installs to roughly 1.2 million. Revenue fell to $367,000 in January, down from December's $540,000. These are not catastrophic numbers in isolation — plenty of apps would accept $367K in monthly consumer spending — but they represent a trajectory that is difficult to reverse without a structural change to how the product reaches users.
For OpenAI, the diagnostic conclusion is uncomfortable but clear: Sora as a destination product — a place people actively choose to go to make videos — has limited sustainable demand at the scale the company needs. The novelty cycle ran its course. People experimented, enjoyed the AI video magic for a few weeks, and then stopped opening the app with enough regularity to sustain the install momentum.
This pattern is not unique to Sora. It mirrors the experience of essentially every dedicated AI media generation app that preceded it: high launch interest, rapid falloff, plateau at a niche-user steady state. The difference is that OpenAI has an obvious escape route that most AI video startups lack — a platform with hundreds of millions of daily users that already has users' trust, payment information, and daily habits baked in.
On March 11, 2026, The Information reported that OpenAI is actively planning to integrate Sora's video generation capability directly into ChatGPT. The integration would allow ChatGPT users to generate AI video without leaving the product they already use daily — no separate download, no separate account, no separate habit to build.
Critically, The Information reported that OpenAI intends to keep the standalone Sora app running even after the ChatGPT integration ships. This is a classic platform strategy: maintain the dedicated product for power users and enthusiasts who want the full Sora experience, while simultaneously embedding the core capability into the platform with the largest user base.
OpenAI had not officially confirmed the plan as of publication. No specific timeline for the integration has been announced. No details about access tiers — whether Sora video generation inside ChatGPT would require a Plus or Pro subscription, or be available at the free tier with usage limits — have been disclosed.
The absence of official confirmation matters less than the strategic inevitability of the move. The standalone app's numbers create a business-case argument that is difficult to ignore: integration into ChatGPT eliminates the user acquisition problem, the habit-formation problem, and the distribution problem in a single decision.
Understanding why OpenAI is pivoting requires understanding what Sora was supposed to be.
When OpenAI first demonstrated Sora in early 2024, it shocked the AI world with photorealistic minute-long videos generated from text prompts. The demos were so impressive that many observers assumed the full product launch was imminent. It took until October 2025 — more than 18 months — for Sora to ship as an accessible consumer product, by which time Google, Runway, and others had advanced their own video generation capabilities considerably.
The October 2025 launch generated the kind of first-week numbers that most apps can only dream of. The 100,000 day-one downloads came despite requiring invites. The Android version, which arrived a few weeks later in November 2025, saw nearly 500,000 installs on its first day according to TechCrunch. Cumulative downloads hit 1 million faster than ChatGPT itself had achieved the milestone.
But the launch was also turbulent. OpenAI's initial approach allowed users to generate videos featuring copyrighted characters — SpongeBob, Pikachu, and similar IP — without restriction, operating on an opt-out model that immediately drew backlash from Hollywood studios and rights holders. By November 2025, OpenAI had pivoted to a stricter opt-in model and tightened content restrictions. The policy change reduced the most viral use cases precisely as the initial novelty wave was cresting. Celebrity deepfake concerns led to particularly strict moderation on content featuring real people, including tighter protocols for historical figures after users generated videos depicting Martin Luther King Jr. in disrespectful contexts.
The combination of policy tightening, reduced novelty, and the inherently episodic nature of video creation — people don't make AI videos every day the way they send text messages — produced the December and January declines that now define the product's standalone trajectory.
If there was a single business event that should have arrested the January decline and didn't, it was the OpenAI-Disney partnership.
The deal — structured as a three-year licensing agreement with Disney making a $1 billion equity investment in OpenAI — gave Sora access to more than 200 characters from Disney, Marvel, Pixar, and Star Wars franchises. Users would be able to generate short, fan-inspired videos featuring licensed characters, under guardrails that prevent inappropriate depictions and exclude actor likenesses and voices.
On paper, this was exactly the kind of content partnership designed to solve Sora's IP problem and give users a reason to return to the app regularly. Mickey Mouse, Iron Man, Baby Yoda — these are the characters people actually want to make videos with, and Disney had officially licensed them.
The Disney partnership launched just as the January decline data was being measured. The fact that installs still fell 45% in that month — with the Disney content available — is the most important data point in this story. It confirms that the decline is structural, not a content-catalog problem. Users who were going to be interested in Sora's capabilities were already interested. The marginal user the Disney partnership was designed to reach didn't download the app even with the world's most recognized character library available.
This is what made the integration decision more urgent. A standalone app with declining installs, an expensive content partnership, and a user acquisition ceiling is not a sustainable business configuration. Moving the capability into ChatGPT is not an admission of failure — it is a recognition that the standalone app was the wrong delivery mechanism for mass market video generation from the start.
The strategic logic of integrating Sora into ChatGPT is almost embarrassingly simple once you accept that standalone video apps face structural user acquisition limits.
ChatGPT had 900 million weekly active users as of February 2026. Monthly traffic to ChatGPT reached 5.72 billion visits in January 2026. These are numbers that no standalone creative AI app can approach, regardless of how good the product is. The distribution advantage is not incremental — it is categorical.
When Sora is a feature inside ChatGPT rather than a separate app, it inherits all of that distribution. Users who are already in ChatGPT composing a blog post, drafting an email, or asking a question can type "make me a short video about this" without context-switching to a different app. The friction reduction is enormous.
More importantly, integration changes the discovery dynamic. Instead of requiring users to know about Sora, seek it out, download it, and build a new habit around it, integration means Sora is available whenever the moment arises organically within ChatGPT's existing workflows. A user editing a social media caption might realize they want a video to accompany it. A user researching a topic might want a visualization. These are moments where video generation adds value but would never trigger an app download — and they happen constantly inside ChatGPT.
The analogy is DALL-E's integration into ChatGPT. OpenAI's image generation product had a standalone presence — users could access it directly — but the integration into ChatGPT dramatically expanded who used it and how often. The same dynamic almost certainly holds for video generation, except the usage lift will be larger because ChatGPT's user base has grown substantially since the DALL-E integration.
The integration is not without significant operational risk. Video generation is substantially more computationally expensive than text generation, and modestly more expensive than image generation. Exposing Sora to ChatGPT's full 900 million weekly active users without careful access controls could produce infrastructure costs that overwhelm any revenue benefit.
WinBuzzer noted this as a central tension in the integration plan. A single AI video generation request at Sora's quality level requires compute resources many times greater than a text chat response. Scaled to even a fraction of ChatGPT's daily query volume — OpenAI processes over 2 billion daily queries — the costs could be prohibitive if video generation is treated as a freely available feature.
The most likely access model mirrors how OpenAI has handled other computationally expensive features: tiered availability. Sora video generation inside ChatGPT would almost certainly be restricted to Plus, Team, or Pro subscribers, at least at launch. Free tier access, if it comes at all, would likely be heavily rate-limited. The standalone Sora app, with its existing subscription revenue, gives OpenAI data on what users will pay for video generation — information that will directly inform how the integrated feature is priced.
There is also a longer-term cost curve to consider. Video generation compute costs have been falling as hardware improves and model efficiency increases. The integration timeline that makes sense from a cost perspective in late 2026 might have been impractical a year earlier. OpenAI is making this move at a moment when the economics are becoming more tractable, even if the cost challenge remains real.
Cost is the tractable challenge. Safety at ChatGPT's scale is the harder one.
Deepfake misuse with Sora's standalone app affected a user base in the low millions. Deepfake misuse with Sora embedded in ChatGPT could affect a platform with 900 million weekly active users and over 2 billion daily queries. The proportional exposure increase is roughly two to three orders of magnitude.
OpenAI has already strengthened Sora's protections following celebrity deepfake concerns, and the Disney partnership required explicit guardrails against inappropriate character depictions. But the moderation infrastructure that works at standalone-app scale is not automatically sufficient at ChatGPT scale.
The specific risks are familiar: generating realistic video of real people saying things they never said, producing content that mimics news footage, creating videos designed to deceive at moments of political or social significance. These are not hypothetical scenarios — they are documented outcomes from less capable video tools that reached mass audiences before adequate moderation systems were in place.
OpenAI's response to this challenge will be as important as the integration itself. The company will need to demonstrate that Sora inside ChatGPT has robust, real-time content moderation that can function at platform scale. The reputational cost of a major deepfake incident on ChatGPT would significantly outweigh the strategic benefit of the video generation integration. Getting the safety infrastructure right before the integration goes broadly live is the correct priority even if it delays the timeline.
The decision to integrate Sora into ChatGPT happens against a video generation competitive environment that has become significantly more crowded since Sora first demonstrated in 2024.
Google Veo 3.1 is currently the benchmark leader in overall preference testing, outperforming Sora 2, Runway Gen 4, and other competitors on the MovieGenBench dataset. Veo 3.1 generates 4K video up to 2 minutes with native audio — including music, sound effects, and dialogue — giving it a meaningful capability advantage over Sora 2, which tops out at 20 seconds of generated video. Google distributes Veo through Gemini and through its Flow AI creative studio, giving it its own embedded-in-platform distribution advantage.
Runway Gen-4.5 leads benchmark Elo ratings at 1,247, ahead of Veo 3 at 1,226 and Sora 2 Pro at 1,206, according to comparative benchmark data. Runway raised $315 million to pursue "world models" that go beyond video clip generation toward persistent simulated environments. It has also strategically integrated Google's Veo 3 and 3.1 directly into its platform, giving subscribers access to multiple top-tier models through a single interface.
Kling AI and Stability AI round out the competitive field, with Kling in particular gaining traction in Asian markets. Audio generation — native sound and music in generated video — is emerging as a key differentiator, with Runway, Kling, and others expected to announce audio features in Q1–Q2 2026.
In this environment, Sora's competitive position improves meaningfully with ChatGPT integration. Not because the underlying model gets better, but because distribution becomes the moat. The best video generator that users never try is less valuable than a good-enough video generator embedded in the app they open dozens of times a day.
For existing Sora standalone app users: Nothing changes immediately. OpenAI has signaled the standalone app will continue operating alongside any ChatGPT integration. Power users and video-first workflows will still have access to the dedicated Sora experience, with its full feature set and creator-focused interface.
For ChatGPT Plus and Pro subscribers: If The Information's report proves accurate, Sora video generation will eventually appear as a native capability inside ChatGPT. Based on OpenAI's pattern with DALL-E image generation, expect it to appear as a mode or tool within the chat interface — type a prompt, select video generation, receive the output without leaving the conversation. Access tier details remain unconfirmed.
For ChatGPT free tier users: Rate-limited access is possible at some point, but gated access to computationally expensive features is OpenAI's standard approach. Free tier video generation, if it comes at all, will almost certainly have strict monthly usage caps.
For developers using the Sora API: The integration into ChatGPT does not directly affect API access. OpenAI's developer products and its consumer ChatGPT product operate on different access paths. What may change is that increased ChatGPT-driven demand for Sora infrastructure could affect API availability and pricing over time.
For the AI video generation market broadly: Sora embedded in ChatGPT normalizes video generation as a routine AI capability rather than a specialty tool. This is the same effect ChatGPT's image generation had on the image AI market — it expanded the addressable user population and raised baseline consumer expectations for what AI should be able to do. That normalization benefits the overall market even as it increases competitive pressure on pure-play video generation products.
No timeline has been confirmed. The Information reported the plans on March 11, 2026, but OpenAI has not made an official announcement. Based on OpenAI's recent product cadence, an integration could arrive anywhere from Q2 to Q4 2026.
Not according to The Information's reporting. OpenAI is expected to maintain the Sora standalone app even after integrating the capability into ChatGPT. The two products serve different use cases: the standalone app for dedicated video creation workflows, and the ChatGPT integration for users who want video generation as one capability among many.
Multiple factors compounded. OpenAI tightened content restrictions in November 2025, reducing the most viral use cases. The initial novelty cycle ran its course. The Disney partnership, while significant, launched too close to the January measurement period to show a meaningful lift. And fundamentally, video creation is an episodic activity — users don't return daily the way they do with chat tools.
Not as a business deal. Disney's $1 billion equity investment in OpenAI and the three-year licensing agreement for 200+ characters is a major partnership. But it did not arrest the January install decline, which confirms the standalone app's growth challenge is structural rather than a content-catalog problem.
OpenAI has not specified. Based on the DALL-E integration pattern, expect core video generation from text prompts as the baseline, with more advanced Sora features potentially restricted to higher subscription tiers or the standalone app.
Google Veo 3.1 currently leads on overall quality benchmarks, including native audio generation and longer video duration (up to 2 minutes vs. Sora's 20 seconds). Sora 2 Pro ranks third in Elo benchmark ratings behind Runway Gen-4.5 and Veo 3. The quality gap is real, but Sora's integration into ChatGPT would give it a distribution advantage that pure benchmark comparisons don't capture.
Almost certainly not at launch. Video generation is computationally expensive, and OpenAI gates high-cost features behind subscription tiers. Expect Sora video generation in ChatGPT to require at least a Plus subscription, with possible rate limits even at that tier.
OpenAI strengthened Sora's deepfake restrictions in October 2025 following concerns about celebrity likenesses. The Disney partnership included explicit guardrails against inappropriate character depictions. The bigger challenge is maintaining these protections at ChatGPT's scale — 900 million weekly active users represents a substantially larger moderation surface than the Sora standalone app.
Sora in ChatGPT normalizes video generation as a standard AI capability, which expands the overall market. But it also puts pressure on standalone video AI products by making a good-enough video tool available inside the app most people already use. Runway's response — accumulating enterprise clients, building toward world models, and integrating Google's Veo — positions it as a professional-grade alternative to consumer-oriented ChatGPT video generation.
Yes. The Android version launched in November 2025 and recorded nearly 500,000 installs on its first day, according to TechCrunch reporting. The Android app exists alongside the iOS app as part of the standalone Sora product.
Based on Appfigures data reported by multiple outlets, Sora's monthly installs fell to approximately 1.2 million in January 2026. The 45% figure represents the month-over-month drop from December's already-declining install rate. For context, the app's cumulative total was 9.6 million downloads across iOS and Android by late January.
Distribution. ChatGPT has 900 million weekly active users and 5.72 billion monthly visits. Sora's standalone app had 9.6 million cumulative downloads. No marketing campaign could close that gap. Embedding Sora in ChatGPT gives it instant access to users who already have the app open daily, without requiring them to discover, download, and build a new habit around a separate product.
No. As of March 14, 2026, OpenAI has not issued an official statement confirming The Information's March 11 report. The company declined to comment publicly on the integration plans.
Sora 2 is the model underlying the current Sora app, representing a significant improvement over the original Sora model demonstrated in 2024. It generates video up to 20 seconds in duration and supports advanced prompt following, camera controls, and scene consistency. Sora 2 Pro is the highest-tier version available in the standalone app.
Enterprise and Teams customers would likely get Sora video generation as part of their existing ChatGPT Enterprise or Team subscriptions, subject to data privacy and security commitments those plans carry. This could be significant for marketing and content teams already using ChatGPT Enterprise who currently use separate video tools.
For certain categories — short social media clips, explainer videos, product demonstrations, and rough-cut visualizations — AI video generation is already capable enough to replace some production workflows. Integration into ChatGPT accelerates adoption in these categories by reducing the friction of accessing the tool. Full replacement of high-production commercial or narrative video is not near-term.
Sora launched into a world where AI was already mainstream, where OpenAI had enormous brand recognition, and where pent-up demand for AI video generation had been building since the original 2024 demos. ChatGPT launched in November 2022 into a market where AI chatbots were still a novelty requiring explanation. The install-pace comparison reflects how much the AI consumer landscape changed in three years, not any performance difference between the products.
The Information is one of the most reliable sources for internal OpenAI developments, with a consistent track record of reporting that is subsequently confirmed. Their March 11 report on the Sora-ChatGPT integration is consistent with the strategic logic and the public data on Sora's install trajectory.
Sora's standalone consumer revenue ($367K in January 2026) would likely decline if users can access comparable video generation inside their existing ChatGPT subscription. But OpenAI would recapture that revenue — and potentially much more — through ChatGPT subscription revenue from users who upgrade or maintain subscriptions partly for video generation access.
Unknown. OpenAI could integrate a full-capability version of Sora into ChatGPT, or a lighter version optimized for the platform's scale and cost constraints. The DALL-E integration into ChatGPT offered comparable quality to the standalone product, suggesting OpenAI tends toward parity — but video generation's computational demands may require more careful quality-cost balancing than image generation required.
The Senate Sergeant at Arms authorized three AI chatbots for staff use while notably excluding Anthropic's Claude and xAI's Grok, raising questions about whether the Pentagon dispute is spreading to Capitol Hill.
OpenAI expands Codex to Windows with doubled rate limits while discontinuing the entire GPT-5.1 model family, signaling an aggressive push toward its next generation.
OpenAI launches Frontier, a new platform for coordinating teams of AI agents across complex enterprise workflows, targeting a market projected to reach $263B by 2035.