# Alibaba Unveils Its Largest AI Model Yet As China Closes The Gap - Forbes

*Источник: Forbes*
*Дата: 2026-08-03*
*Язык: en*

**Кратко:** Alibaba on Monday released Qwen3.8-Max, calling it the largest and most capable model in its Qwen AI series so far, and investors responded immediately: the stock rose 4.5% in premarket trading in New York and 7% in Hong Kong following the announcement. The company says the model can go toe-to-toe with America's best.

Alibaba on Monday released Qwen3.8-Max, calling it the largest and most capable model in its Qwen AI series so far, and investors responded immediately: the stock rose 4.5% in premarket trading in New York and 7% in Hong Kong following the announcement.
The company says the model can go toe-to-toe with America's best. Alibaba shared benchmark results showing Qwen3.8-Max delivering comparable — and occasionally better — scores than Anthropic's Claude Fable 5, the model widely regarded as the current frontier. According to the company, the model outperformed leading U.S. systems on several coding, multimodal and engineering benchmarks while trailing on some general-purpose reasoning tests. On independent leaderboards, the picture is more nuanced: Qwen3.8-Max immediately became the highest-ranking Chinese model for text tasks on the crowdsourced Arena.AI platform, though it still trails several Anthropic offerings, and it ranked second globally for vision tasks, behind only a Fable 5 variant.
The technical specifications are striking. The multimodal model has 2.4 trillion total parameters and supports a context window of up to 1 million tokens, and it uses a sparse mixture-of-experts architecture that activates only 95 billion parameters per operation, reducing computing costs and response times compared with dense models of similar scale. Alibaba is also making a notable strategic move: Qwen3.8-Max will become the first Max-class Qwen model to be open-sourced, with weights available for public download next week — a return to open-sourcing top-tier models after the company kept several recent flagship releases proprietary earlier this year.
The company is pitching the model at long-horizon autonomous work. In one internal test, Alibaba said, the model spent 16 days building and improving an AI coding tool — writing code, testing it, fixing errors and refining its work with little human input.
A Crowded Month For Chinese AI
The release lands in the middle of an extraordinary run for Chinese labs. Moonshot AI unveiled Kimi K3 at the World Artificial Intelligence Conference in Shanghai on July 17 — a 2.8 trillion parameter mixture-of-experts model with a one million token context window — and demand was so intense that the Beijing startup temporarily paused new subscriptions, saying demand had pushed close to the limits of its capacity and that its GPUs were "feeling it." Bloomberg reported that Qwen3.8-Max ranks higher than Kimi K3 on some benchmarks.
Taken together with recent releases from DeepSeek and ByteDance, the pattern is hard to miss. Vey-Sern Ling of Union Bancaire Privée says the gap between Chinese and American AI is "narrowing fast." Other analysts have put numbers on that assessment: Wei Sun, principal AI analyst at Counterpoint Research, estimates the general gap between Chinese and American models has narrowed to three to six months, though it varies significantly by task.
Wall Street sees a business story as much as a technology one. Citi analysts noted that rapid, frequent AI model releases are pushing enterprises toward a "model-agnostic" approach — picking the best model and price per task rather than committing to a single provider. That dynamic favors challengers, and Chinese labs are pressing the advantage on price and openness: while leading American developers have increasingly favored closed systems, Chinese companies have become prominent suppliers of open-weight models that can be downloaded and adapted by developers worldwide.
The Caveats
The claims deserve scrutiny. Vendor benchmarks arrive faster than independent verification, and Alibaba has not yet published a full benchmark table, model card or license for the release. There are practical limits, too: a 2.4 trillion parameter model occupies well over a terabyte even after heavy compression, keeping it firmly in data-center territory. And the compute crunch that forced Moonshot to turn away paying customers illustrates a structural constraint — Chinese AI labs continue to face computing power shortages as U.S. export controls restrict access to advanced chips and chipmaking equipment.
The next test comes quickly. The open-weights release is scheduled for next week, at which point independent researchers can verify Alibaba’s claims for themselves. Whatever those tests show, the cadence itself is the story: China's top labs are now shipping frontier-scale models weeks apart, and the market is repricing accordingly.

[Оригинал](https://www.forbes.com/sites/gabrielalinzainescu/2026/08/03/alibaba-unveils-its-largest-ai-model-yet-as-china-closes-the-gap/)