StepFun has no news.

"Journalism at Its Finest"

"AI products launch daily, foundation models update monthly. Benchmarks and slide decks everywhere, invite codes and agents flying in tandem. Yet nobody cares except AI content creators.

And does anyone even remember, before the Hangzhou Six Little Dragons, the previous generation's gods were the AI Six Little Tigers 😭

In light of this, I hereby declare this week as the Savage Roast Week. Roasting each of the AI Six Little Tigers one by one. Keeping it real.

One post per day, and today is the finale: StepFun."

My biggest question about StepFun is: who put them on the AI Six Little Tigers list?

StepFun has no news. My only impression of StepFun is: Shanghai's little Zhipu AI. Beijing has Zhipu AI, so Shanghai needed StepFun.

I asked several friends and couldn't dig up a single piece of gossip. The only thing that came to mind is that StepFun shut down its consumer app Maopaoya, and tried to spin off its application team Lipu, but hasn't completed the spin-off yet.

This company seems to have entered a state of being forgotten ahead of schedule.

Do you still remember the previous generation's AI Four Little Dragons? Besides SenseTime and Megvii, there was also CloudWalk Technology. StepFun's current situation is exactly that vibe. So the critical question is: who exactly put StepFun in the Six Little Tigers, and why not choose ModelBest instead?

Pure journalism at its finest, friends 😭

Now look at the entire AI Six Little Tigers, and you'll discover an even bigger problem.

Excluding 01.AI, which is already completely out of the game, the remaining companies have collectively fallen into a middling, neither-here-nor-there predicament.

The fundamental contradiction lies in valuation.

The AI Six Little Tigers have mostly raised several hundred million dollars, with valuations of $2-3 billion. Having taken this much money, you're obligated to tell a platform-level, foundation-model-building story.

But reality is harsh.

On one hand, the foundation models aren't happening. These companies have released a bunch of models over the past six months, but not a single one has impressed anyone.

When users actually want a top-tier model, they still go to ChatGPT, Claude, and Gemini. For domestic models, the Qwen ecosystem is rich and diverse, and everyone's waiting for DeepSeek's new release. Apart from AI content creators, nobody cares what the Six Little Tigers launched.

On the other hand, going all-in on applications can't sustain their valuations either. A pure application company simply can't support a valuation of tens of billions of RMB. There's no way to face investors.

So the collective problem of the Six Little Tigers is: too high to reach, too low to settle.

They have to pretend they're still grinding away at foundation models, while secretly building applications to find a way out. This "want both" state creates internal chaos at these companies.

This is a structural contradiction.

To find a way out, companies inevitably need to experiment with applications. But because they're carrying high valuations, their KPI expectations for applications are absurdly high — say you built an AI-native application, and the first thing the boss considers is whether it can hit ten million DAU.

But at this point in 2025, AI applications are still in the small-product, small-tool stage, suitable for indie developers and small teams to tinker with. Nobody can see the mature form of AI products.

At this moment in time, you simply can't build an AI application with ten million DAU. Product managers can't convince their bosses.

So the people building applications at the Six Little Tigers have it especially rough.

A product manager works hard to build a somewhat interesting small application that shows some traction. The boss looks at DAU, sees it's nowhere near the ten-million target, and immediately kills the project.

This is a structural contradiction, and the direct consequence is that product leaders at the Six Little Tigers are leaving in batches.

These product managers go out to start their own companies, no longer burdened by billion-RMB valuations, raising ten million dollars at $100-200 million valuations, focusing solely on building one AI application. They end up living more comfortably.

So all this chaos stems from the awkward situation of being held hostage by high valuations.

And the starting point of this situation was the AI Six Little Tigers list created by media and investors.

So who says journalism doesn't matter?

The AI Six Little Tigers are journalism at its finest. Entrepreneurs still need to get along with media professionals. Next time they compile the "AI Seven Little Dragons" or "AI Eight Vajras," try to get yourself included, and don't follow in ModelBest's footsteps 😭

(Article illustrations generated by ChatGPT o3, with writing assistance from an AI product currently in development.)