During the National Day holiday, I saw several KOLs in the cryptocurrency field praising Everlyn AI on Twitter—a "video generation" project similar to Sora. This piqued my curiosity and prompted me to delve deeper into my research.
My first impression was quite stunning: an AI project led by former Meta research scientists, claiming to be building a groundbreaking open-source AI video generation model and constructing a decentralized Web3 protocol. Its crypto narrative integrates multiple hot sectors such as AI, intellectual property (similar to Story Protocol), DePIN (io.net is also an investor), and the creator economy. When a project combines numerous hot topics and has endorsements from well-known figures, it is indeed impressive.
However, after further investigation, I found myself deeply skeptical of this project. Perhaps due to my limited information sources, I could not find any substantial trace of this project in the Web3 world. Simply put, the project lacks publicly verifiable protocol addresses and blockchain explorers to observe its on-chain activities. Aside from the genuine issuance of its tokens, the other Web3 elements seem as elusive as "air."
This article will provide a comprehensive analysis of Everlyn AI based on publicly available information, helping readers understand the true nature of this project.
Overhyped AI and Illusory Blockchain Promises
Core Claims of the Project
The core selling points that Everlyn AI promotes include:
In terms of AI
Everlyn-1 Video Generation Model: Claimed to be the "first open-source autoregressive video model," its frame-by-frame video generation method is similar to GPT's "next token prediction." This contrasts with the slower diffusion models commonly used by competitors. Academic research confirms that autoregressive (AR) models have potential in decoding speed and generating long sequences but are prone to error accumulation; while diffusion models excel in generation quality but are computationally expensive. Everlyn chooses the autoregressive route, aiming for breakthroughs in speed and length, claiming to generate 1080p resolution, 8-second videos in 4 seconds while reducing costs by tenfold.
Open-source Commitment: A core differentiating feature of Everlyn AI is its "open-source" nature, promising to provide "fully open-source model weights," directly challenging the industry status quo of closed-source models like OpenAI's Sora.
In the Web3 Field
Descriptions of Everlyn's innovative highlights in AI are relatively consistent in existing materials. However, descriptions of its Web3 characteristics vary widely, with some content even missing from official documents. I summarize its Web3 narrative as follows:
With the proliferation of deepfake technology, the need to verify the authenticity of digital content has become unprecedentedly urgent. Everlyn claims to be building a dedicated Layer 1 public chain that records timestamps and creator information on-chain, providing immutable source proof and copyright verification for AI-generated content. This way, videos created through Everlyn will leave a publicly transparent record on the blockchain, making it easy to identify and trace if used to create deepfake content. This narrative bears similarities to the logic of IP platforms like Story Protocol and NFTs.
Everlyn has built a DePIN platform where users can contribute their GPUs to provide computing power for "video generation" and earn rewards. This narrative may be related to io.net's investment involvement, thus sparking market associations with its DePIN narrative.
Additionally, it includes the more common creator economy narrative, where creators can receive token incentives for publishing videos on the Everlyn platform.
These claims accurately target current market pain points: high costs of AI video generation, monopolization by closed-source models, and difficulty in verifying content authenticity. However, there is a significant gap between promotion and reality.
AI Component: Real but Overhyped
Despite the strong team background, there are clear contradictions in its technical promotion:
Terminology Confusion: The project claims to use an "autoregressive model" architecture but mentions "xDiT (Distributed Diffusion Transformer)" in promotional materials. The diffusion transformer is a core component of diffusion models, which are two different technical routes from autoregressive models. This confusion of fundamental concepts is concerning.
Code Repository Status: Although there is an Everlyn-Labs organization on GitHub, its repository mainly contains academic research projects (such as ANTRP, Wasserstein-VQ) published by team members, not a unified, production-level video generation system. Furthermore, there is no blockchain-related code in the team's GitHub repository.
Lack of Performance Validation: The project claims to be "the fastest video generator on Earth," but lacks independent third-party benchmark validation. In the open-source community, models like CogVideoX and CausVid have gained widespread recognition, while the actual competitiveness of Everlyn-1 has yet to be proven.
Web3 Component: Nonexistent "Ghost"
This is the most serious issue with the entire project. The project claims to build a decentralized video AI layer, yet after thorough searching, I could not find any relevant Web3 technical white papers, developer documentation, public testnets, blockchain explorers to observe on-chain activities, or any code repositories related to Web3 technology????.
In summary, Everlyn's Web3 component, aside from the genuine existence of token issuance (its token is on BSC, with the address 0x302DFaF2CDbE51a18d97186A7384e87CF599877D), everything else remains at the conceptual narrative level, completely lacking technical details.
With the absence of code implementation, its Web3 part has only hollow narratives (it is worth mentioning that Everlyn's official Web3 narrative is relatively restrained, and many eye-catching claims come from some KOLs' overinterpretations), yet it is eager to issue tokens. This behavior pattern gives me a familiar "pump and dump" vibe.
The Truth Behind Celebrity Endorsements
The credibility of a project largely depends on its founding team. I verified the backgrounds of the two co-founders, Dr. Harry Yang (Co-founder & CTO) and Dr. Ser-Nam Lim (Co-founder & Research Lead/CEO), and found no issues. Both indeed worked at Meta and have made significant contributions in the field of image AI.
Aside from the two founders, all materials praising the project highlight that Turing Award winner and Meta's Chief AI Scientist Yann LeCun is an advisor to Everlyn AI as a key selling point to prove the project's academic achievements. However, based on my investigation, LeCun's "advisor" status seems inflated.
The only official information source supporting this claim is a tweet from the Everlyn team stating, "We’re honoured to welcome Yann LeCun as an academic advisor to Everlyn." However, I checked LeCun's personal website (yann.lecun.com), Meta AI page (ai.meta.com), LinkedIn profile, and all relevant posts on his X account, and found no mention of Everlyn AI or his advisory role. Moreover, LeCun's posts and interviews mainly discuss general AI topics, with little reference to Web3, let alone the Everlyn project.
Therefore, I believe that this association with LeCun is a typical case of "celebrity effect" marketing—magnifying a trivial interaction to create the illusion of a formal endorsement relationship.
The Players Behind: Investment Logic and Marketing Machine
Strategic Investment from Mysten Labs
Everlyn AI secured $15 million in funding led by the Sui development team, Mysten Labs, with a valuation of $250 million. However, it is noteworthy that neither Mysten Labs' official blog nor Sui Foundation channels have released any official statement regarding this investment.
This "silence" suggests to me that Mysten Labs' investment is at best a strategic talent acquisition—to acquire a top AI team and its technology to enrich the Sui ecosystem, rather than endorsing the project's token economics or decentralized roadmap.
Interpreting the Real Signal of "Listing on Binance"
Being listed on Binance is also a highlight of the project's promotion, but we need to accurately understand the real signal behind it:
The LYN token is listed on the "Binance Alpha" platform, not the main spot trading area of Binance.
Binance Alpha is a high-risk sandbox specifically for trading "emerging digital assets not listed on Binance."
Binance explicitly warns: being marked as an Alpha asset does not mean it will be listed on the main site in the future, and the asset faces higher volatility and risk, potentially losing all investments and being non-withdrawable.
Therefore, being listed on the Alpha platform does not serve as a quality endorsement but rather resembles a trading experiment conducted under the premise of isolating risk.
Kaito-Driven Marketing Machine
The astonishing popularity of Everlyn AI is not coincidental; it is backed by a systematic marketing promotion machine.
Role of Kaito: Everlyn AI used Kaito's Capital Launchpad platform for its $2 million public offering. Kaito claims to be an "AI-driven vertical search engine," and its launch platform allocates quotas based on user social reputation and other metrics. Notably, Everlyn AI's airdrop tasks explicitly require users to post in Kaito's "yaps" section.
Paid Promotion Mechanism: Kaito's business model includes a "Yap-to-Earn" competition, where project parties set up reward pools to "incentivize creators to produce content about the company." This directly confirms my suspicion: the support from numerous KOLs for Everlyn is a result of paid promotion.
The conclusion is evident: the massive market presence of Everlyn AI does not stem from the organic appeal of its technology or spontaneous enthusiasm from the community, but rather from a costly, meticulously planned marketing campaign through professional platforms like Kaito. Linking posting ("yapping") directly to rewards (points, airdrop weight) creates a clear economic incentive for KOLs to promote the project without needing to conduct in-depth verification of its quality. The high "heat index" on platforms like RootData is a result of this artificially generated activity, rather than a true reflection of the project's fundamental progress.
Conclusion
In summary, I personally believe that Everlyn AI, a recent hot project heavily promoted by many KOLs, is extremely suspicious and warrants caution. The founders attract capital using their AI reputation and collaborate with crypto marketing teams to issue tokens. The AI component may be real, but its performance as advertised is still difficult to verify. However, based on my observations, its Web3 component is entirely a "vapor project," a "ghost project"—it has only narrative without code, yet is eager to issue tokens, presenting a troubling eagerness to "pump and dump."
The Everlyn AI case reveals a concerning phenomenon in the cryptocurrency field: using legitimate technical teams and endorsements from top investors to provide credibility cover for token projects lacking substance. This "decentralized drama" is more deceptive than a pure scam because some of its parts (AI research) are real, but the investment vehicle (tokens and so-called protocols) is built on a fictional foundation.
For investors, the core lesson is: do not relax scrutiny of a project's infrastructure just because the team is excellent or KOLs are promoting it. In the Web3 field, the absence of fundamental elements such as technical white papers, testnets, and open-source code is a more important danger signal than any marketing promotion.
This article is based on publicly available information and does not constitute investment advice. Cryptocurrency investment carries significant risks; please make cautious decisions and DYOR.
If you liked this article, feel free to follow, like, and share your support!
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
