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

marsbitPublished on 2026-03-18Last updated on 2026-03-18

Abstract

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.

Apple Finally Admits, Siri Is Getting Old

In a significant shift, Apple has rebranded Siri to "Siri AI" at WWDC 2026, acknowledging the assistant's limitations after years of stagnation. The company announced a deep partnership with Google, leveraging Gemini's model capabilities to train its new Apple Foundation Models. This collaboration extends Apple's Private Cloud Compute to Google Cloud and Nvidia GPUs for the first time. The article traces Siri's history from its groundbreaking 2011 debut to its subsequent confinement within Apple's closed ecosystem, prioritizing control and privacy over expansive functionality. While Apple integrated AI into its hardware and systems over the years (e.g., Neural Engine, Core ML), it missed the paradigm shift brought by generative AI models like ChatGPT. Facing pressure, Apple restructured its AI leadership and opted to license Google's Gemini technology—reportedly paying around $1 billion annually—to power the revamped Siri. The strategy involves "distilling" knowledge from the large Gemini model into smaller, on-device models. Apple also plans to use Google Cloud's Nvidia GPUs for complex cloud inference tasks. The core vision for "Apple Intelligence" is a system-level assistant that reduces cognitive load: summarizing notifications and emails, drafting context-aware replies, and retrieving relevant information across apps. Siri gains a dedicated app with memory and cross-device sync. However, this AI push comes with hardware requirements, potentially excluding older iPhones. A major challenge is China, where Apple Intelligence will likely be a different product due to local regulations, requiring partnership with a domestic AI provider. The article concludes by questioning the future of personal AI, noting that true understanding involves more than data access—it requires knowing where to stop. Apple's partnership marks a humble beginning in its quest to build a genuinely helpful, yet respectful, personal assistant.

marsbit3m ago

Apple Finally Admits, Siri Is Getting Old

marsbit3m ago

How Far is the Current U.S. Stock Bull Market from Historical Bubble Peaks?

Title: How Far Is the Current U.S. Stock Bull Market from Historical Bubble Peaks? Summary: According to Goldman Sachs' chief U.S. equity strategist, current market exuberance ranks at the 86th historical percentile. While approaching, this level remains notably below the extremes seen at the peaks of the 2000 dot-com bubble (100th percentile) and the 2021 bull market (95th percentile). The recent 15% surge in the S&P 500 over two months represents one of the strongest volatility-adjusted rallies in over 50 years, primarily fueled by AI-related enthusiasm and backed by significant upward revisions to earnings estimates, providing a more fundamental basis than past momentum-driven peaks. Key risk signals—speculative frenzy, deteriorating growth, heavy equity issuance, and Federal Reserve tightening—are all closer to their historical triggers than a few months ago but are not yet fully activated. For instance, IPO activity is recovering, and market breadth remains narrow, though not at dot-com bubble extremes. The strategist cautions that while the current cycle may not replicate past peaks exactly and a downturn does not require extreme euphoria, risks are accumulating. The overall assessment is that the bull market still has room to run, but the window of opportunity is gradually narrowing.

marsbit11m ago

Rapper Afroman: Doesn't Understand Bitcoin, but Becomes a New Crypto Freedom Fighter

Joseph "Afroman" Foreman, the rapper famed for "Because I Got High," has become an unlikely symbol for the crypto community's libertarian ideals. Following a 2022 police raid on his Ohio home, which yielded no charges, Foreman used the footage to create viral songs mocking the officers, leading to a lawsuit he ultimately won. His defense of free speech against government overreach resonated deeply with attendees at the Bitcoin 2024 conference in Las Vegas, where he performed. Despite admitting he knows little about cryptocurrency and owns none, event organizers saw his personal fight as embodying Bitcoin's core ethos of individual sovereignty and resistance to centralized control. Foreman, a self-described "icon of weed culture," has long performed for paying crowds but found a new political voice after his legal battle. His platform, which includes federal marijuana legalization and police accountability, aligns with the diverse, anti-establishment crowd at the Bitcoin conference, though he remains unaffiliated with any major party. The event featured an auction of his memorabilia, celebrating him as a modern "freedom fighter." In a final surprise, Foreman used his stage time to announce his candidacy for the 2028 U.S. presidential election, thrilling the remaining audience before launching into his signature hit.

Foresight News15m ago

Rapper Afroman: Doesn't Understand Bitcoin, but Becomes a New Crypto Freedom Fighter

Foresight News15m ago

When Doing Cryptocurrency Payment, the First Thing is Licenses, What is the Second?

When launching a crypto payment business, obtaining the necessary licenses is the crucial first step. However, the second, and arguably more critical, step is designing a comprehensive operational framework that forms a coherent business loop. This loop must be clearly understood and executable by all stakeholders: banks, payment partners, exchanges, on-chain analytics providers, regulators, and your internal team. Many projects mistakenly believe a single license permits all operations. Licenses merely grant entry; they don't define how the specific business functions. The real challenge lies in detailing every aspect of the workflow. This involves clarifying the customer base, the flow of fiat and crypto assets, the settlement process, and establishing clear lines of responsibility for risks like AML compliance, sanctions screening, chargebacks, and regulatory inquiries. A robust framework must answer seven core questions: Who are the clients and merchants? Who collects fiat and crypto? Who handles conversion and custody? And who is ultimately accountable for compliance and risk management? Projects often fail not from a lack of licensing, but during due diligence when they cannot convincingly explain these operational details. Therefore, beyond securing licenses, the priority must be constructing a closed-loop system. This system ensures the business model is transparent, risks are managed, responsibilities are delineated, contracts are aligned, and the entire process is comprehensible to partners and regulators. The true competitive edge in crypto payments lies not in acquiring a license quickly, but in integrating licensing, banking, compliance, and operations into a sustainable and executable whole.

marsbit49m ago

When Doing Cryptocurrency Payment, the First Thing is Licenses, What is the Second?

marsbit49m ago

Arthur Hayes Analysis: AI Bubble Nears Burst, Crypto Market Faces Short-Term Pressure

Arthur Hayes argues that the current AI market is a bubble poised to burst, which will exert downward pressure on the crypto market in the near term. The core trigger is rising oil prices due to the US-Iran conflict and a blockade of the Strait of Hormuz. Higher energy costs directly increase the operational expenses of AI data centers, squeezing profit margins for companies like Google, Anthropic, and OpenAI. Hayes predicts that persistent inflation from high oil prices will force Trump, in a bid to win the November election, to turn public sentiment against the AI industry. He may propose regulations and taxes on data centers and AI companies to appeal to voters concerned about costs and job displacement. Such political rhetoric could shatter market confidence. Furthermore, the market is unlikely to healthily absorb the massive concurrent IPOs of SpaceX, Anthropic, and OpenAI, which together seek valuations in the trillions. The combination of soaring energy costs, overwhelming equity supply, and negative political pressure will puncture the AI bubble. Hayes notes that nearly all new USD liquidity since 2022 has flowed into AI, leaving crypto like Bitcoin behind. When the AI bubble bursts, liquidity will contract sharply, pulling down all risk assets, including cryptocurrencies. In response, Hayes's fund, Maelstrom, has sold all AI-related stocks and non-core cryptocurrencies. It maintains core positions in Bitcoin and Ethereum while increasing exposure to energy sector equities, betting on rising oil and gas prices. He expects Bitcoin to bottom after the AI-led market decline, before rallying again with future monetary easing.

Foresight News1h ago

Arthur Hayes Analysis: AI Bubble Nears Burst, Crypto Market Faces Short-Term Pressure

Foresight News1h ago

Trading

Spot

Futures

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

Abstract

Related Questions

Related Reads