🛎️ MiniMax Shakes X Overnight

Good Morning, AI Enthusiasts!

I was half asleep when the barrage of comments on X woke me up: yet another Chinese giant model had joined the fray.

NEW LAUNCH

MiniMax M2.5 Release Shakes X Overnight

👀 What’s happening: MiniMax released M2.5 in the early hours, positioning it as an open source frontier model built for real productivity. Within minutes, benchmark screenshots spread across X. Posts citing 80.2 percent on SWE Bench Verified and strong search and agent scores flooded timelines. In a few hours, the conversation passed 1.8 million views and dominated tech feeds.

🌍 How this hits reality: The reaction was not just about performance. It was about price and speed landing at the same time. M2.5 claims 37 percent faster execution and pricing near one dollar per hour at 100 tps. That shifts the economics of long running agents for startups watching burn rates and API bills more closely than ever.

🛎️ Key takeaway: If the numbers hold in production, agent deployment moves from experiment to baseline infrastructure. The real race shifts to orchestration, distribution, and who captures developer mindshare while attention is still hot.

TOGETHER WITH QODER

Free on Qoder — With the New MiniMax-M2.5

MiniMax-M2.5 just launched — the fastest agentic model among mainstream LLMs. Built natively for agent workflows, it runs 3x faster than Opus at a fraction of the cost.

Qoder is offering 3 days of free access to celebrate the launch. No credit card required.

If you're building AI agents or want to test something beyond Claude, this is your window.

CODING

Musk Says Intent Will End Coding

👀 What’s happening: Elon Musk says programming dies this year. Not evolves. Dies. By December, AI will soon skip coding languages and compilers and generate machine instructions directly. No writing syntax. No debugging scripts. Just describe what you want and the system builds it. Linked with Neuralink, he calls it imagination to software.

🌍 How this hits reality: Software runs in layers. Humans write Python or C++, which compiles into the machine instructions CPUs actually execute. Tools like Lovable still sit inside that stack, generating traditional code under the hood, yet they have already accelerated delivery by orders of magnitude. If AI skips the human readable layer entirely and moves straight from intent to optimized machine instructions, the compression would be extreme. What feels fast now would look primitive.

🛎️ Key takeaway: If intent replaces syntax, the leverage shifts from engineers to those who define problems clearly. The bottleneck moves upstream. AI infra and chip efficiency become the real choke points, not coding talent.

TOGETHER WITH MYCLAW

MiniMax M2.5 is Already Live on MyClaw

If you’ve been watching the benchmarks on X, now is the time to test it in the real world.

MyClaw.ai lets you safely run OpenClaw with new MiniMax M2.5 model in the cloud that stays online 24/7. No local setup. No fragile environments. Just plug in, deploy your agent, and see how M2.5 performs on actual workflows.

Coding, search, long running tasks. Try it under real load, not just screenshots.

Spin up your OpenClaw with M2.5 on MyClaw and experience the difference yourself.

MISTRAL

Europe Turns Mistral Into Strategic AI Hedge

👀 What’s happening: Mistral’s annualized revenue surged from $20 million to over $400 million within a year, with a stated goal of reaching $1 billion in recurring revenue by year end. More than 100 enterprise clients, including ASML and TotalEnergies, now rely on it. Around 60 percent of revenue comes from Europe as local governments and boardrooms push for digital sovereignty.

🌍 How this hits reality: The EU still sources over 80 percent of its digital services from foreign, mostly American providers. That dependency is now viewed as geopolitical exposure. Mistral is committing 1.2 billion euros to Swedish data centers with EcoDataCenter, starting with a 23 megawatt site. Infrastructure runs customer workloads by day and trains models at night. This is a real try at sovereign vertical integration, not just scale.

🛎️ Key takeaway: If procurement shifts from performance to jurisdiction, regional AI stacks gain structural advantage. Expect more capital into local compute, tighter data residency rules, and mounting pressure on US hyperscalers across Europe.

PENTAGON

Pentagon Forces AI Into Classified Tradeoff Zone

👀 What’s happening: The Pentagon is pressing OpenAI, Anthropic, Google and xAI to deploy frontier models onto classified military networks, with fewer built in restrictions than civilian users face. Officials say they want AI operating across all classification levels, including mission planning and potentially weapons related systems.

🌍 How this hits reality: Classified networks handle targeting data, operational plans and sensitive intelligence. Generative models can synthesize massive data flows faster than human teams, which matters when millions of defense personnel already use AI on unclassified systems. But model errors inside mission environments carry consequences measured in lives, not bug reports.

🛎️ Key takeaway: This is a dual constraint game. The military wants speed and dominance. AI firms want boundaries and control. Whichever side yields defines how frontier models integrate into sovereign war infrastructure.

DAILY TL;DR

OpenClaw Trending

• Legendary said on X that China’s newly released GLM-5 and Kimi K2.5 frontier models are running in his workflows via OpenClaw, and he compared the two.
• Kyohei Ogawa suggested OpenClaw users buy a DockKit-compatible charger and give it a “face” and “body” to create a more companion-like presence.
• OpenClaw has surpassed VSCode in GitHub stars and joked it could become an AI-powered one-person unicorn.
• Frank said that the real gold rush begins when even the “dumbest guy you know” is making six to seven figures from OpenClaw wrappers.

AI Trending

• A Malaysian man bought the domain AI.com in 1993 for about $3–4 and sold it in 2025 for roughly $75–80 million.
• Blackstone increased its investment in Anthropic to about $1 billion at a $350 billion valuation.
• ByteDance launched its video AI model Seedance 2.0, which went viral in China and drew comparisons to DeepSeek.
• As OpenAI moves toward ads in ChatGPT, Anthropic expanded Claude’s free features, including file editing and Skills, to attract users.
• India requires platforms to label all AI-generated content by February 20, pressuring Instagram and X despite immature deepfake detection tools.

TRENDING

Daily AI Launches

• Tines offers a secure, vendor-agnostic platform for building and monitoring intelligent automation workflows.
• Migma AI has launched an AI email platform with one-click domain setup, API access, and precise conversion tracking.
• Subscription Day² has launched on iOS, bringing its redesigned Mac subscription tracking tools to the iPhone.
• Revo AI Email Assistant has launched a tool that connects Slack, CRM, and meetings to auto-answer emails in Gmail and Outlook.

Featured AI Tools

• ✨ MuseMail.ai makes creating stunning, on-brand emails from one prompt with AI precision.
• 🪶 CopyOwl is the First AI Research Agent, deep research on any topic in one click.
• 🦾 Flot AI writes, reads, and remembers across any apps and webs.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next
• Oncely helps you discover the next wave of AI products and launches

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.