What this is

A developer used the same prompt to have 9 popular AI tools build an identical 8th-grade physics PPT on buoyancy. The lineup: Doubao, Ake AI, Kimi, DeepSeek, Xiaomi MiMo, ChatGPT, WorkBuddy, Gamma, and Canva.Final ranking: Ake AI > Doubao > Gamma, ChatGPT > Canva > DeepSeek, MiMo. Kimi hit a paywall queue; WorkBuddy's run was voided after a one-hour wait. Neither completed the test.Two anomalies worth flagging: first, the little-known Ake AI beat every major player; second, DeepSeek and Xiaomi MiMo—two Chinese general-purpose LLMs (foundation models that handle a wide range of tasks)—could only output HTML rather than deliver an actual PPT file.

What the industry thinks

The optimistic read: Ake AI, a vertical product, carving out space in a market dominated by ChatGPT and Gamma shows there's still room at the AI application layer. Founders don't have to bet everything on foundation models.But we care more about three problems this benchmark exposes. First, the sample is one PPT with one prompt; "Ake AI is #1" carries limited weight—closer to a "small-sample champion." Second, ChatGPT produced overlapping text, Canva's image placement was off, DeepSeek literally delivered HTML—proof that even big-name general models fail at something as small as a PPT without vertical tuning. Third, when 9 tools do the same thing and the gap between them is thin, the space is saturated, and the next round of culling is coming fast.A colder read: DeepSeek and MiMo—LLMs that can "write code and chat"—losing to an unknown small vendor on something as trivial as a PPT is the most direct evidence that "general capability ≠ scenario delivery."

What it means for regular people

For enterprise IT: Pick AI tools by delivered output, not brand pedigree. Chinese tools like Doubao and Ake AI already cover basic PPT needs—no need to worship overseas names.For individual professionals: If free tools work, don't pay for subscriptions. Overseas tools like Canva adapt poorly to Chinese teaching content; localization remains the moat for domestic tools.For the consumer market: AI tools have shifted from "can it do it?" to "does it do it well?" Every player is iterating fast. Today's rankings will likely reshuffle within six months.