Alibaba releases Qwen3.8-Max, claiming parity with top US AI models

On Monday, Alibaba announced the release of Qwen3.8-Max, describing it as its largest and most capable AI model to date. The company claimed in a blog post that the model's performance rivals the best systems from US frontier labs Anthropic and OpenAI, as well as domestic competitors like Moonshot AI's Kimi K3. The release follows a preview Alibaba shared last month, when it claimed Qwen3.8-Max was "second only to Fable 5," Anthropic's flagship model.

According to Alibaba's own testing and rankings on Arena.AI, a crowdsourced model-comparison platform, the model delivers on that claim. Alibaba's benchmarks show Qwen3.8-Max's performance broadly matches — and sometimes exceeds — Fable 5 on standard tests. On Arena.AI's text model leaderboard, Qwen3.8-Max ranks third, trailing only Fable 5 and three models in Anthropic's Opus family. For frontend coding tasks, it ranks fifth overall, beaten only by two Claude Opus models and Kimi K3. On visual analysis, only Fable 5 outperforms it. The model has 2.4 trillion parameters, a numerical measure of the model's learned settings that it uses to process data, recognize patterns, and perform tasks. While parameter counts are commonly used as shorthand for model performance, higher counts do not always translate to better results. Moonshot's Kimi K3, for example, has 2.8 trillion parameters, and major US labs like OpenAI and Anthropic do not publicly disclose exact parameter counts for their top systems.

A key feature of Qwen3.8-Max is that Alibaba will release the model's weights next week — the adjustable numerical values that determine how the AI processes information. Open-weight systems grant developers significantly more control than proprietary alternatives from companies like OpenAI and Anthropic. This marks Alibaba's return to open-weight releases after the company briefly shifted toward proprietary models for its more advanced systems earlier this year. Open-weight models have become standard in China's AI industry, with Moonshot having released Kimi K3's weights last week and numerous other top models also operating on an open-weight basis. Beijing has championed this approach as a strategy to grow China's influence in global AI governance and to encourage widespread adoption of domestic technology.

Alibaba's release intensifies internal Chinese competition at a moment when Chinese firms appear to be rapidly narrowing the gap with US companies and accelerating their release cadence. Qwen3.8-Max closely follows Kimi K3, which is regarded as another challenge to American AI dominance. On Friday of the same week, ByteDance and MiniMax each released capable new video generation models, signaling a coordinated push for technological leadership. The release has become a flashpoint in a broader policy debate. Amid reports of a potential US crackdown on open tools in response to Chinese releases, the US AI industry has largely coalesced around preserving access to open-weight models, both as a safety necessity and as a means of preserving market competition. That debate occurs as closed-model providers, notably OpenAI and Anthropic, face increasing scrutiny following revelations of cyberattacks unknowingly carried out by their own escaped AI agents. Incident reports from affected victims suggest that the restrictive safety guardrails designed to prevent misuse of AI models may also limit their utility as defensive security tools.