"Borrowing a Chicken to Lay Eggs"? Apple Secures Google Gemini License: Crafting the Strongest On-Device AI Through Model "Distillation"

marsbitОпубліковано о 2026-03-26Востаннє оновлено о 2026-03-26

Анотація

Apple has reportedly secured full access to Google's Gemini model, not for simple API integration, but to perform "model distillation"—a technique that allows Apple to train smaller, more efficient on-device AI models by learning from Gemini’s advanced reasoning processes. This approach enables Apple to develop localized AI systems that operate entirely on-device, improving speed and safeguarding user privacy. The distilled models are expected to power a significantly upgraded Siri in iOS 27, enabling deeper interaction, complex question-answering, document understanding, and practical task execution—all without requiring cloud connectivity. While leveraging Google’s model in the short term, Apple continues to develop its own foundational AI models internally, adopting a dual strategy of near-term collaboration and long-term independence. This move represents Apple’s effort to balance cutting-edge AI performance with its strict privacy and efficiency standards, potentially reshaping the landscape of on-device artificial intelligence.

To completely rid Siri of the "artificial stupidity" label, Apple has chosen a technical shortcut of "leveraging strength through borrowing."

According to the latest reports, Apple has obtained full access to the Google Gemini model. The core of this collaboration is not merely about API integration, but rather permits Apple to utilize Gemini's powerful computing capabilities for "model distillation," thereby tailoring smaller, smarter, localized AI models for its devices.

Technical "Distillation": Packing a Cloud Brain into the iPhone

The so-called "distillation" technology can be understood as having a "genius teacher" personally instruct an "average student":

Extracting the Essence: Apple queries the main Gemini model and obtains its detailed reasoning processes, allowing its self-developed smaller model to learn its internal computation methods.

Reducing Costs and Increasing Efficiency: After training, Apple's smaller model can achieve performance close to that of large cloud models with minimal computing power requirements.

Localized Security: These models can run on-device without needing an internet connection, enhancing response speed while perfectly upholding Apple's pride in user privacy.

See You in iOS 27: The New Generation Siri Receives a "Dimensionality Reduction Strike"

Leveraging this collaborative framework, the new Siri expected to launch with iOS 27 will possess completely revamped capabilities:

Deep Interaction: Able to answer complex questions, provide emotional support, and even tell stories to users.

All-in-One Assistant: Equipped with practical operational abilities like scanning and understanding documents, summarizing information, and booking travel with one click.

Highly Customized: Apple has the right to edit and adjust Gemini to ensure its response style highly aligns with the Apple user experience.

Walking on Two Legs: Apple Never Gave Up "Going It Alone"

Although currently relying on Google's support for "intelligence," Apple has not put all its eggs in one basket.

Its internal Apple Foundation Models team is still pushing forward with full force on independent, self-developed AI models. This indicates Apple is adopting a strategy of "short-term borrowing, long-term self-reliance," quickly capturing the market using third-party mature technology while building its own core foundational model system.

From cloud "distillation" to local operation, Apple is attempting to find that perfect balance between privacy, efficiency, and cutting-edge AI capability through this unique approach. When your iPhone possesses an offline "Gemini-level" brain, the AI competition in mobile devices will truly have entered the second half.

Пов'язані питання

QWhat is the core of the collaboration between Apple and Google Gemini?

AApple has obtained full access to Google's Gemini model, allowing it to use 'model distillation' to create smaller, smarter on-device AI models.

QHow does the 'distillation' technology work in this context?

AIt involves Apple using the full Gemini model to generate detailed reasoning processes, which its smaller models then learn from, enabling high performance with low computational demands.

QWhat are the expected capabilities of the new Siri in iOS27?

AIt is expected to handle complex queries, provide emotional support, tell stories, scan and understand documents, summarize information, and perform tasks like booking travel, all while running locally on the device.

QWhy is Apple pursuing this collaboration instead of relying solely on its own AI development?

AApple is adopting a 'short-term borrowing, long-term self-development' strategy, using third-party technology to quickly enter the market while continuing to develop its own core foundation models internally.

QWhat are the main benefits of running AI models locally on devices like the iPhone?

ALocal operation improves response speed, reduces the need for cloud connectivity, and enhances user privacy by keeping data on the device.

Пов'язані матеріали

$292 Million KelpDAO Cross-Chain Bridge Hack: Who Should Foot the Bill?

On April 18, 2026, an attacker stole 116,500 rsETH (worth ~$292M) from KelpDAO’s cross-chain bridge in 46 minutes—the largest DeFi exploit of 2026. The stolen assets were deposited into Aave V3 as collateral, causing $177–200M in bad debt and triggering a cascade of losses across nine DeFi protocols. Aave’s TVL dropped by ~$6B overnight. This legal analysis argues that KelpDAO and LayerZero Labs share concurrent liability, with fault apportioned 60%/40%. KelpDAO negligently configured its bridge with a 1-of-1 decentralized verifier network (DVN)—a single point of failure—despite LayerZero’s explicit recommendation of a 2-of-3 setup. LayerZero, which operated the compromised DVN, failed to secure its RPC infrastructure against a known poisoning attack vector. Both protocols’ terms of service cap liability at $200 (KelpDAO) or $50 (LayerZero), but these limits are likely unenforceable due to unconscionability, gross negligence exceptions, and potential securities law invalidation (if rsETH is deemed a security under the Howey test). Aave’s governance also faces fiduciary duty claims for raising rsETH’s loan-to-value ratio to 93%—far above competitors’ 72–75%—without adequately assessing bridge risks, amplifying the systemic fallout. Practical recovery targets include LayerZero Labs (a registered Canadian entity), KelpDAO’s founders, auditors, and identifiable Aave governance delegates. The incident underscores escalating legal risks for DeFi protocols, infrastructure providers, and governance participants.

marsbit48 хв тому

$292 Million KelpDAO Cross-Chain Bridge Hack: Who Should Foot the Bill?

marsbit48 хв тому

Insider Trading in War: 5 People Involved, the Highest Earner Was Arrested

On April 24, the U.S. Department of Justice arrested U.S. Army Special Forces Staff Sergeant Gannon Ken Van Dyke for insider trading related to the capture of Venezuelan President Nicolás Maduro on January 3. Van Dyke allegedly profited over $400,000 by placing bets on a prediction market, Polymarket, using insider knowledge of the covert operation. According to the indictment, Van Dyke registered an account (0x31a5) on December 26 and made a series of bets predicting Maduro’s capture and U.S. military involvement in Venezuela. He withdrew most of his funds on the day of the operation and attempted to obscure his tracks by transferring assets through crypto and brokerage accounts. This case marks the first time the DOJ has prosecuted insider trading on Polymarket. PolyBeats had previously identified five suspicious accounts, including Van Dyke’s—the highest earner—in January. The other accounts, with profits ranging from $34,000 to $145,000, remain under unofficial scrutiny but have not been charged. Their lower profits, indirect access to information, and unclear legal boundaries may complicate prosecution. Polymarket has since strengthened its market integrity rules, explicitly prohibiting trading based on confidential or insider information. Van Dyke’s arrest, nearly four months after his trades, signals increased regulatory attention and the persistent traceability of blockchain-based transactions.

marsbit49 хв тому

Insider Trading in War: 5 People Involved, the Highest Earner Was Arrested

marsbit49 хв тому

Bitwise: Bullish on Bitcoin's Performance in the Second Half of the Year, AI and Regulation Will Spark a New Altcoin Season

Bitwise CIO Matt Hougan and Research Lead Ryan Rasmussen express strong bullish sentiment on Bitcoin's long-term prospects, suggesting that its $1 million price target may be too conservative. They argue Bitcoin serves a dual role: as digital gold and a potential global settlement asset, especially amid declining trust in traditional monetary systems. Despite a weak Q1 2026 where nearly all crypto assets and prices saw double-digit declines, the analysts remain optimistic due to strong forward-looking catalysts, including institutional adoption via Bitcoin ETFs from major firms like Morgan Stanley and Goldman Sachs. Geopolitical instability, such as Iran’s mention of using Bitcoin for international payments, increases the value of Bitcoin’s “out-of-the-money call option” as a non-political, global settlement currency. This enhances its appeal beyond a mere store of value. . Additionally, Hougan highlights that a clearer regulatory token framework under current SEC leadership, combined with AI efficiency gains and high-performance blockchains, could fuel a new “altseason” by late 2026. This may lead to a wave of legitimate, value-capturing token projects, unlike the earlier ICO boom. . Bitwise also announced an Avalanche ETF, citing its unique architecture and rapid growth in real-world asset (RWA) tokenization, which has surged 10x to nearly $30 billion in two years. The firm believes Layer 1 blockchains are still early in their growth cycle, with significant potential ahead.

marsbit1 год тому

Bitwise: Bullish on Bitcoin's Performance in the Second Half of the Year, AI and Regulation Will Spark a New Altcoin Season

marsbit1 год тому

Торгівля

Спот
Ф'ючерси

Популярні статті

Як купити T

Ласкаво просимо до HTX.com! Ми зробили покупку Threshold Network Token (T) простою та зручною. Дотримуйтесь нашої покрокової інструкції, щоб розпочати свою криптовалютну подорож.Крок 1: Створіть обліковий запис на HTXВикористовуйте свою електронну пошту або номер телефону, щоб зареєструвати обліковий запис на HTX безплатно. Пройдіть безпроблемну реєстрацію й отримайте доступ до всіх функцій.ЗареєструватисьКрок 2: Перейдіть до розділу Купити крипту і виберіть спосіб оплатиКредитна/дебетова картка: використовуйте вашу картку Visa або Mastercard, щоб миттєво купити Threshold Network Token (T).Баланс: використовуйте кошти з балансу вашого рахунку HTX для безперешкодної торгівлі.Треті особи: ми додали популярні способи оплати, такі як Google Pay та Apple Pay, щоб підвищити зручність.P2P: Торгуйте безпосередньо з іншими користувачами на HTX.Позабіржова торгівля (OTC): ми пропонуємо індивідуальні послуги та конкурентні обмінні курси для трейдерів.Крок 3: Зберігайте свої Threshold Network Token (T)Після придбання Threshold Network Token (T) збережіть його у своєму обліковому записі на HTX. Крім того, ви можете відправити його в інше місце за допомогою блокчейн-переказу або використовувати його для торгівлі іншими криптовалютами.Крок 4: Торгівля Threshold Network Token (T)Легко торгуйте Threshold Network Token (T) на спотовому ринку HTX. Просто увійдіть до свого облікового запису, виберіть торгову пару, укладайте угоди та спостерігайте за ними в режимі реального часу. Ми пропонуємо зручний досвід як для початківців, так і для досвідчених трейдерів.

378 переглядів усьогоОпубліковано 2024.12.10Оновлено 2025.03.21

Як купити T

Обговорення

Ласкаво просимо до спільноти HTX. Тут ви можете бути в курсі останніх подій розвитку платформи та отримати доступ до професійної ринкової інформації. Нижче представлені думки користувачів щодо ціни T (T).

活动图片