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.

Is Anyone Still Buying in the Crypto Market? Unpacking 3 Common Watch-and-Wait Mentalities Today

Is Anyone Still Buying in the Crypto Market? Unpacking 3 Common Wait-and-See Mindsets This article analyzes the current cautious sentiment in the crypto market, distilled from conversations with sophisticated investors. The author identifies three dominant investor mindsets: 1. **Satisfied with Current Holdings:** Many retain a long-term belief in digital assets but see no immediate catalyst for significant price appreciation. They hold positions to avoid missing a future surge but allocate minimal new capital or attention. A shift requires a new, observable catalyst or a rotation from other portfolio areas. 2. **Waiting for Lower Prices:** This reflects not just short-term timing but a belief about crypto's total addressable market and upside potential. It could change if key perceived cycle bottoms pass without a crash, a major bullish event occurs (e.g., sovereign adoption), or price rebounds trigger FOMO-driven buying. 3. **High Opportunity Cost of Allocation:** The core question is comparative growth. With AI-related equities appearing to offer relentless, high-speed growth, justifying marginal investment into assets without similar perceived momentum is difficult. A slowdown in the AI trade could potentially mark a bottom and trigger capital reallocation into crypto. In conclusion, while long-term conviction persists for many, near-term marginal capital flows are constrained by these beliefs. The author suggests the market may be closer to a bottom than a top, but the current climate is defined by this wait-and-see approach, awaiting a catalyst to reignite broader investor commitment.

marsbit8m ago

Is Anyone Still Buying in the Crypto Market? Unpacking 3 Common Watch-and-Wait Mentalities Today

marsbit8m ago

China Added 67 New Unicorns in Half a Year, with AI and Robotics Accounting for Over Half

China added 67 new unicorn companies in the first half of 2026, reaching a total of 517 unicorns with a combined valuation of approximately $2.39 trillion. This surge marks a significant rebound after a post-2022 slowdown and sets a new semi-annual record. The growth is primarily driven by Artificial Intelligence (AI) and Robotics, which together account for over 53% of the new entrants. Specifically, 19 new unicorns are in robotics and 17 in AI. Notable companies include DeepSeek ($615.38B) and Kling AI ($18B). The trend indicates a decisive shift from internet consumer models to hard tech innovation. Geographically, new unicorns are highly concentrated in four cities: Beijing (19), Shanghai (18), Shenzhen (9), and Hangzhou (5), which together host 76.1% of the new companies. Hangzhou's overall valuation is boosted significantly by DeepSeek. Valuation distribution among new unicorns is pyramidal: 77.6% are valued between $1B and $2B, indicating early-stage status, while only two exceed $10B. There is a notable "speed divide": many AI/robotics startups achieved unicorn status in under three years, often via corporate spin-offs or led by star founders, while hard tech companies in semiconductors or biotech typically took over eight years. The report concludes that this wave reflects China's accelerating transition into an AI and robotics-powered innovation cycle, characterized by faster company formation, heightened geographic concentration, and a clear focus on foundational technologies.

marsbit51m ago

China Added 67 New Unicorns in Half a Year, with AI and Robotics Accounting for Over Half

marsbit51m ago

Vibe Coding Accidentally Burns $550,000 Worth of Tokens, Then Turns Mockery into $700 Million Profit

An employee at the American fintech company Slash, encouraged by his company to experiment with AI "vibe coding," created a simple pixel-style FPS browser game called "Brainrot Shooter." He initially shocked the online community by posting a fake screenshot showing an exorbitant $81,267 (550,000 RMB) AI token bill for developing the game. The story quickly went viral, with many expecting him to be fired for the massive "expense incident." However, the narrative took a dramatic turn. The game attracted significant attention, garnering thousands of players and hundreds of hours of playtime within 48 hours, and even drew interest from brands for advertising partnerships. Slash reframed the project as a successful "strategic initiative." The employee later announced that the viral game had indirectly brought in approximately $100 million in new Assets Under Management (AUM) for Slash's platform—a claimed 1200x return. Finally, the employee revealed the truth: the original $81k bill was fabricated using browser developer tools; the actual development cost was less than $200. The entire episode, from the fake bill to the viral success and massive business impact, was an elaborate and highly effective marketing stunt.

marsbit1h ago

Vibe Coding Accidentally Burns $550,000 Worth of Tokens, Then Turns Mockery into $700 Million Profit

marsbit1h ago

Aave V4 crosses $250M – But ONE liquidity challenge still remains

Aave V4 has surpassed $250 million in deposits, marking significant early adoption of its upgraded lending infrastructure. This growth is driven by enhanced capital efficiency, risk parameters, and lending options. However, a portion of these deposits came from users migrating from V3, meaning not all represent new capital entering the ecosystem. For sustained growth, Aave must continue attracting net new liquidity. While V4 deposit numbers are strong, the broader Total Value Locked (TVL) tells a more nuanced story. After a market downturn, overall TVL has recovered to 7.4 million ETH but remains below its all-time high, with withdrawals still limiting net liquidity growth. A positive sign is the substantial increase in cbETH deposits, which rose from around $18-$20 million in May to approximately $70 million by early July. This indicates growing demand for liquid staking collateral and strengthens Aave's lending capacity. Aave's future position as a dominant DeFi liquidity provider hinges on its ability to secure consistent, net-positive capital inflows beyond internal migrations.

ambcrypto1h ago

Aave V4 crosses $250M – But ONE liquidity challenge still remains

ambcrypto1h ago

Probability Falls Below 50%: Is the Clarity Act Dead This Year?

The article discusses the prospects of the U.S. "Clarity Act," a comprehensive federal regulatory framework for digital assets. Originally aiming for a July 4th signing, that deadline has been missed, and with only about three weeks of effective legislative work before the August recess, its passage this year is uncertain. Key points include: * The bill, which aims to clarify regulatory boundaries between the SEC and CFTC, passed the House in July 2025 and a Senate committee in May 2026. * Several unresolved issues are hindering progress, including disagreements over stablecoin revenue provisions, liability exemptions for DeFi developers, and ethics and enforcement details. A recent closed-door negotiation on ethics provisions broke down in June. * A positive development is that the Major County Sheriffs of America (MCSA) has shifted its stance on the bill from opposition to neutral, citing addressed concerns over a clause related to developer liability. * According to Polymarket data, the probability of the Clarity Act being signed into law this year is currently only 49%, reflecting cautious market expectations. * Analysts note that passage could accelerate institutional adoption of crypto, while further delays would prolong regulatory uncertainty.

Foresight News2h ago

Probability Falls Below 50%: Is the Clarity Act Dead This Year?

Foresight News2h ago

Trading

Spot