Alibaba Says Its New Qwen Model Is 'Second Only to Fable 5' — but Shared No Benchmarks

Alibaba has staked a ranking claim in the AI race: it says its new Qwen3.8-Max-Preview is "one of the most powerful models available today, second only to Fable 5." The claim came from the company's own post on X. But as SCMP points out, something is missing: Alibaba didn't say which test or which score that ranking is based on.

In brief

• What launched: Qwen3.8-Max-Preview, a preview of Alibaba's strongest Qwen model, at 2.4 trillion parameters.
• The claim: in Alibaba's own words, "one of the most powerful models available today, comparable to leading frontier AI models, second only to Fable 5."
• What's missing: no benchmark name or score was given to support it, and there's no independent verification. For now it's a qualitative statement.
• Access: via a Token Plan subscription and the Qoder and QoderWork agentic platforms. Pricing wasn't disclosed, and it wasn't described as open-source.

Our take

The value here for you isn't the model — it's the claim itself. When a company says "we're second in the world," the first question should be: on which test, with what score, measured by whom? Alibaba gave none of the three, which leaves just a sentence. Most announcements we've seen in recent weeks came with at least a benchmark table; some companies (like Moonshot) even wrote down their own shortcomings. So treat this less as model news and more as a reading habit: when picking AI, look at the measurable number and who produced it, not the shiny sentence. And if the model does turn out to be great, you lose nothing — you just waited for the evidence.

Kaynak: SCMP