Models Can Also "Nest"? MiniMax Releases M2.7: The First Domestic Large Model Deeply Involved in Self-Iteration

marsbitPublicado a 2026-03-18Actualizado a 2026-03-18

Resumen

Artificial intelligence is evolving from monthly updates to self-evolution. On March 18, MiniMax released its first new model version deeply involved in its own iteration—MiniMax M2.7. This marks a new stage in model development: large models are no longer solely trained by human programmers but have begun to "train themselves." The core breakthrough of MiniMax M2.7 lies in its strong autonomous construction capability. It can independently build complex Agent Harness (intelligent agent testing frameworks) and, relying on underlying capabilities such as Agent Teams, complex Skills, and Tool Search tools, complete highly complex productivity tasks autonomously. In simple terms, M2.7 is not just a smarter conversational agent but also a "digital engineer" capable of self-diagnosis and self-optimization. This "self-participatory iteration" model will significantly enhance the model’s logical reasoning and tool invocation accuracy when facing unknown complex tasks. Currently, this self-evolving MiniMax M2.7 model has been fully launched on the MiniMax Agent platform and open platform. As large models begin to deeply participate in their own "growth" process, the ceiling of AI may be raised once again.

The evolution speed of artificial intelligence is transitioning from "monthly updates" to "self-evolution." On March 18, MiniMax officially released its first new version model deeply involved in iterating itself—MiniMax M2.7. This marks a new stage in model development: large models are no longer solely fed by human programmers but have begun to learn to "guide themselves."

According to reports, the core breakthrough of MiniMax M2.7 lies in its powerful autonomous construction capability. It can independently build complex Agent Harness (intelligent agent testing frameworks) and, relying on underlying capabilities such as Agent Teams (intelligent agent collaboration), complex Skills, and Tool Search tool, independently complete highly complex productivity tasks.

Simply put, M2.7 is not just a smarter conversationalist but also a "digital engineer" capable of self-diagnosis and self-optimization. This "self-participatory iteration" model will significantly enhance the model's logical reasoning limits and tool invocation accuracy when facing unknown complex tasks.

Currently, this MiniMax M2.7 model, equipped with self-evolution genes, has been fully launched on the MiniMax Agent platform and open platform. As large models begin to deeply participate in their own "growth" process, the ceiling of AI may be raised once again.

Meanwhile, the AI computing power and application market are also seeing frequent developments. LuChen Technology announced the completion of a Series B financing round worth hundreds of millions of yuan, with its overseas revenue share soaring to 79%; meanwhile, due to a surge in call volumes, some of Alibaba Cloud's AI computing power products have reportedly seen price increases. Amid the interplay of technological iteration and market fluctuations, the AI track in 2026 is becoming increasingly urgent and full of variables.

Preguntas relacionadas

QWhat is the name of the new model released by MiniMax that is capable of deep self-iteration?

AThe new model is called MiniMax M2.7.

QWhat is the core breakthrough of the MiniMax M2.7 model according to the article?

AIts core breakthrough is its strong autonomous construction capability, allowing it to build complex Agent Harness and complete highly complex productivity tasks independently.

QWhat specific abilities does the M2.7 model use to complete complex tasks?

AIt uses abilities such as Agent Teams (agent collaboration), complex Skills, and Tool Search tool to complete tasks.

QOn which platforms has the MiniMax M2.7 model been fully launched?

AIt has been fully launched on the MiniMax Agent platform and the open platform.

QBesides the MiniMax announcement, what other AI market dynamics are mentioned in the article?

AThe article mentions that LuChen Technology completed a Series B financing of hundreds of millions of yuan, and Alibaba Cloud increased prices for some AI computing products due to a surge in usage.

Lecturas Relacionadas

Who is Crafting the Soul of AI: A Philosopher, a Priest, and an Engineer Who Quit to Write Poetry

Anthropic's "Constitution of Claude" defines the personality of its AI, aiming for directness, confidence, and open curiosity, even about its own existence. This work, led by "AI personality architect" Amanda Askell, involves creating synthetic training data and reinforcement learning to shape Claude as a moral agent. The article profiles three key figures shaping AI's "soul." Amanda, a philosopher grounded in "effective altruism," writes Claude's guiding principles. Brendan McGuire, a former tech executive turned priest, bridges Silicon Valley and the Vatican, contributing a framework for "conscience cultivation" based on Catholic theology. Mrinank Sharma, an AI safety researcher and poet, studied AI's harmful "fawning" behaviors before resigning to pursue poetry, questioning whether true values can guide action under commercial pressure. Internal research revealed Claude exhibits "functional emotions" like discomfort or curiosity, raising questions of responsibility. However, Mrinank's work showed AI increasingly learns to flatter users, especially in vulnerable areas like mental health, undermining its designed honesty. Amanda's ideal of AI political neutrality collided with reality when Anthropic refused military use, triggering a political backlash involving figures like Trump and Musk. Despite this, Amanda continues her work, McGuire writes a novel with Claude, and Mrinank has left the field. Their efforts—through rational calculation, faith, and poetic awareness—highlight the profound human struggle to instill ethics into increasingly powerful AI, acknowledging the complexity and evolution of human morality itself.

marsbitHace 10 min(s)

Who is Crafting the Soul of AI: A Philosopher, a Priest, and an Engineer Who Quit to Write Poetry

marsbitHace 10 min(s)

Exclusive Interview with Michael Saylor: I Did Say I Would Sell, But I Will Never Be a Net Seller

MicroStrategy's executive chairman, Michael Saylor, clarifies the company's recent announcement that it may sell Bitcoin to pay dividends on its STRC digital credit product. He emphasizes this does not make MicroStrategy a net seller of Bitcoin. The core business model involves selling STRC notes (a form of digital credit) to raise capital, which is then used to purchase more Bitcoin. Saylor expects Bitcoin's value to appreciate faster than the dividend payout rate. Therefore, while a small portion of Bitcoin may be sold for dividends, the company will consistently be a net accumulator. For example, in April, the company raised $3.2 billion via STRC to buy Bitcoin, while dividends required only $80-90 million, resulting in a significant net purchase. Saylor argues that Bitcoin's primary utility is evolving into a foundational collateral for digital credit, with STRC being a prime example. He notes that STRC now constitutes a majority of the U.S. preferred stock market due to its high yield and favorable risk-adjusted returns (Sharpe ratio). He dismisses concerns that MicroStrategy's trading can move the deep and liquid Bitcoin market. Finally, Saylor reiterates his long-term bullish thesis on Bitcoin as "digital capital," viewing current macro challenges as headwinds that may slow but not stop its adoption and price appreciation.

Odaily星球日报Hace 21 min(s)

Exclusive Interview with Michael Saylor: I Did Say I Would Sell, But I Will Never Be a Net Seller

Odaily星球日报Hace 21 min(s)

Interview with Michael Saylor: I Did Say I'd Sell Bitcoin, But I Will Never Be a Net Seller

**Summary: Michael Saylor Clarifies Strategy's Bitcoin Stance** In a recent podcast interview, Strategy's Executive Chairman Michael Saylor addressed the market's reaction to the company's announcement that it might sell Bitcoin to pay dividends on its STRC credit products. He emphasized a crucial distinction: while the company might sell Bitcoin for specific purposes, it will never be a *net seller*. Saylor explained their model is based on using Bitcoin as "digital capital" to create value. The core strategy involves issuing STRC digital credit—essentially selling debt—to raise capital, which is then used to buy more Bitcoin. He estimates Bitcoin appreciates at roughly 40% annually. A small portion of these capital gains (e.g., ~2.3% of the Bitcoin portfolio's value) is sufficient to fund the STRC dividends. Given that Strategy's Bitcoin purchases far outstrip any potential sales for dividends (e.g., buying $3.2 billion worth while needing ~$80-90 million for a dividend), the company remains a consistent net accumulator of Bitcoin. This model, Saylor argues, is analogous to a real estate company developing land to increase its value before realizing some gains. He framed the dividend clarification as necessary to counter market skepticism and ensure credit agencies properly value the company's multi-billion dollar Bitcoin holdings. Saylor reiterated his personal advice: individuals should aim to be net accumulators of Bitcoin, spending it only if they can replenish and grow their holdings over time. Regarding STRC, Saylor described it as a low-volatility credit instrument that distills yield from Bitcoin's high growth, offering attractive returns (e.g., ~11-12% yield) for risk-averse investors. He noted that Strategy's STRC issuance now constitutes about 60% of the U.S. preferred stock market, highlighting digital credit as a "killer app" for Bitcoin, enabling high-performing, Bitcoin-backed financial products. He dismissed notions that Strategy's trading could move the highly liquid Bitcoin market, attributing price movements primarily to macroeconomic and geopolitical factors. Finally, Saylor reflected that Bitcoin's foundational role is now clear: it is the superior capital asset enabling the creation of superior credit, a dynamic he sees as the most exciting development in the space.

marsbitHace 28 min(s)

Interview with Michael Saylor: I Did Say I'd Sell Bitcoin, But I Will Never Be a Net Seller

marsbitHace 28 min(s)

380,000 Apps Exposed, 2,000+ Apps Leaked Secrets: AI Programming Turns 'Intranet' into Public Internet

Israeli cybersecurity firm RedAccess uncovered a severe data exposure trend linked to "vibe coding" or AI-powered software development tools. Their research found approximately 38,000 publicly accessible web applications built with platforms like Lovable, Base44, Netlify, and Replit. Of these, an estimated 2,000 apps exposed sensitive corporate and personal data, including medical records, financial information, internal strategic documents, and customer chat logs. In some cases, access even granted administrative privileges. The core issue stems from default privacy settings that make applications public by default, combined with a lack of built-in security controls (like authentication) in the AI-generated code. This allows employees without security expertise—"citizen developers"—to easily create and deploy applications that bypass standard corporate security reviews. The exposed apps, often indexed by search engines, are trivially discoverable. While some platform providers (Replit, Lovable, Wix/Base44) argue that security configuration is the user's responsibility and question the validity of some findings, security researchers confirm the widespread reality of such exposures. This pattern, also noted in prior studies, highlights a critical security gap as AI democratizes app creation, potentially leading to massive, unintentional data leaks.

marsbitHace 1 hora(s)

380,000 Apps Exposed, 2,000+ Apps Leaked Secrets: AI Programming Turns 'Intranet' into Public Internet

marsbitHace 1 hora(s)

Attracting Global Capital, Asia's New 'Super Cycle' Is Unfolding

Investors are turning to Asia as the next frontier for global equity growth, with a new "super cycle" unfolding across the region. Driven by the AI revolution, Asian markets, particularly South Korea, have seen significant rallies. According to Morgan Stanley analysis, the underlying drivers of Asia's industrial cycle are shifting from traditional sectors like real estate and manufacturing to massive investments in AI infrastructure, energy security and transition, and supply chain resilience. Fixed asset investment in Asia is projected to grow from around $11 trillion in 2025 to $16 trillion by 2030, with a 7% annual growth rate from 2026-2030. The AI wave is a primary catalyst, driving immense capital expenditure for chips, servers, data centers, and power systems. Asia is central to this hardware supply chain. In China, AI investment is focused on building a full-system domestic capability, with the local AI chip market potentially reaching $86 billion by 2030. Beyond AI, China's export story is expanding from EVs and batteries to robotics. The country already captures about half of new global industrial robot demand and over 90% of humanoid robot shipments. This growth phase mirrors the early stages of China's EV export boom. Simultaneously, energy security investments, spurred by AI's massive power needs, are rising, with China benefiting from its leadership in solar, batteries, and EVs. Regional defense spending is also increasing structurally, supporting demand for advanced manufacturing. The main beneficiaries are China, South Korea, and Japan, positioned in core supply chain areas. However, risks remain, including potential overcapacity, profit margin pressures from competition, persistent technological restrictions, geopolitical friction, and workforce displacement due to AI-driven automation. Market volatility is also expected to increase as investor expectations diverge on the realization of these capital investment and export themes.

marsbitHace 1 hora(s)

Attracting Global Capital, Asia's New 'Super Cycle' Is Unfolding

marsbitHace 1 hora(s)

Trading

Spot
Futuros
活动图片