Qwen3.8-Max Raises the Stakes

Alibaba is betting on scale and endurance

Sixteen days is the number.

Alibaba says its new Qwen3.8-Max model worked autonomously for that long while building and improving an AI coding tool. The model has 2.4 trillion parameters, supports a 1 million-token context window, and is scheduled for release next week.

The scale will grab attention. The more important claim is that the model can stay useful across long coding, research, legal, and visual tasks with limited human input.

Alibaba’s shares rose 4.5% in New York and 7% in Hong Kong, which shows investors are treating model releases as business signals again. I think the real test will be reliability, because a system working for days can create more value, and much bigger mistakes, than one answering a single prompt.

The pressure is rising quickly.

Moonshot AI has already released a larger 2.8 trillion-parameter model, and Alibaba is framing Qwen3.8-Max against Anthropic’s best systems. What happens when endurance matters more than benchmark rank?

Microsoft wants agent research to be easier to build

The interesting part is the infrastructure.

Microsoft Research has open-sourced Orchard, a framework for building, training, and evaluating AI agents across coding, web browsing, and productivity tasks. Instead of releasing another model, Microsoft is releasing the environment, training workflows, datasets, and evaluation tools that let researchers reuse the same infrastructure across different agent systems.

The benchmark numbers are impressive, including 69.7% on SWE-bench Verified, rising to 73.0% with value-model reranking using roughly 3 billion active parameters. What stands out more is the idea that smaller models can stay competitive when the surrounding system is well designed. The screen in a developer's workspace matters as much as the model running behind it.

This shifts some attention away from model size and toward the software stack around agents. If more labs can train directly inside real deployment environments, progress could spread beyond the handful of companies with proprietary infrastructure.

The next question is whether the open community builds on Orchard fast enough to make it a shared standard.

AI spending is changing the whole economy

The AI story has moved well beyond Silicon Valley.

The latest Wall Street Journal reporting shows AI spending reaching into the broader U.S. economy, with tech companies pouring hundreds of billions of dollars into computing infrastructure and raising billions in debt to fund it. The chart points to investment in software, hardware, communications equipment, and data centers climbing toward $1.4 trillion by 2026.

This feels bigger than another tech investment cycle because the spending is now influencing capital markets, manufacturing, and even consumer prices. You can hear the constant hum of construction around new data centers in many regions, and that tells you these investments are becoming physical, not theoretical.

The payoff still has to justify the cost. If AI delivers meaningful productivity gains, today's spending could look reasonable. If adoption moves more slowly, companies carrying large debt loads may face tougher questions from investors.

The next year matters. The biggest question is whether the economic gains start showing up as clearly as the spending already has.

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.