Kimi’s Innovative Payment Model in AI Products

———— / BEGIN / ————

Recently, many domestic AI products have been exploring payment models, as these types of products consume a significant amount of computing power, and offering them for free is not sustainable in the long term.

Yesterday, while chatting with everyone about AI products, I came across an interesting image of KimiChat.

Before the emergence of AI products, the common payment method was a subscription model, similar to membership on video streaming platforms. However, we find that many AI products have introduced usage limits even within the monthly subscription framework. For example, Midjourney limits the number of images that paid users can generate quickly each month.

The reason lies in the significant difference in cost structure between AI products and the aforementioned video streaming services.

Video platforms incur minimal bandwidth costs after shooting content, so adding new paid members does not significantly increase costs for the company.

In contrast, the cost structure of AI products, particularly generative AI, is largely based on computing power consumption—meaning that the more users utilize the service, the higher the costs incurred by the company, hence the imposition of usage limits.

At this point, we have only discussed the payment method of monthly subscriptions with usage limits.

However, the payment model I saw yesterday for KimiChat was truly eye-opening.

If we look at the image, we see that Kimi’s payment model is described not in terms of paid benefits but rather as a form of “tipping”.

This distinction is quite significant.

Firstly, from the user’s perspective: if I pay under the guise of a purchase and receive corresponding benefits, I will have high expectations for the product’s capabilities and perceive my relationship with the product as transactional rather than emotional.

In contrast, the tipping method creates a very different psychological shift, akin to the feeling of a top donor tipping a streamer or a leader rewarding a subordinate.

The result is that after tipping you, I will consider you one of my own. This strengthens the bond between the user and the product, naturally enhancing user retention. This phenomenon aligns with a fascinating psychological effect.

[The depth of our feelings for someone is not determined by how well they treat us, but rather by how well we treat them.]

Diving deeper into the psychology of tipping reveals another highlight.

Consider this:what does the act of tipping itself represent?

Simply put, it signifies user recognition of the product.

Users tip because they find the product useful and it meets their needs. However, KimiChat has transformed the act of payment into a tipping behavior.

This approach enhances the psychological suggestion of user recognition of the product from the opposite direction.

Nonetheless, the essence remains the same; from the image, we can see that it fundamentally involves purchasing different priority usage days at varying amounts.

In real life, this can be compared to a user buying something that is not very useful but still claiming to others, “This thing is fantastic,” to avoid appearing foolish.

Of course, this is not to say that KimiChat’s product is subpar. My actual experience using it has been quite fine; we are merely analyzing this approach from a psychological perspective.

Having analyzed from the user’s viewpoint, there are also many beneficial aspects from other perspectives.

For instance, when the product design team sees user tips, they feel that their product is recognized by users.

This can motivate the product design team to continue optimizing features and experiences, fostering innovation to provide better services to users. From this perspective, the tipping description can even enhance internal team efficiency.

For the company, having a certain income can support sustainable development, helping the team acquire more resources and avoid solely relying on investor funds, which could lead to bankruptcy once those funds are exhausted.

Additionally, it allows for some preliminary exploration of the commercial model for AI assistant products.

Currently, the majority of internet products primarily generate revenue through advertising; however, introducing ads will inevitably affect user experience, impacting usability, discoverability, attention guidance, and other critical experience metrics.

Furthermore, KimiChat’s design of different tipping amounts also features some eye-catching designs.

We can see that different amounts correspond to varying usage days. Upon closer observation, we notice that the days used are irregular, such as 4 days, 8 days, 23 days, 40 days, etc., rather than the familiar “X amount for one month”.

In my understanding, the irregular days have the advantage of avoiding direct comparison with traditional monthly subscription products. For example, if a user calculates that a month costs 36 yuan, they might think, “I don’t even want to pay 20 for a Youku membership, so I won’t buy this either.”

Another highlight is that they may have employed reverse price sensitivity anchors to boost users’ willingness to pay.

Price anchoring means that if I see a bottle of water priced at 25 yuan next to a coffee priced at 29 yuan at Starbucks, the water price influences my perception of the coffee price, making 29 yuan seem reasonable. In reality, other brands might sell coffee for just 9 yuan, which illustrates the significance of price anchoring.

Carefully calculating the daily usage price from the image shows that the pricing does not offer significant discounts as the number of days increases, remaining around 1.2 to 1.3 yuan per day:

An important piece of information is that when comparing the price of purchasing 365 days against 93 days, it may even turn out that buying 93 days is cheaper on a daily average.

So, why is purchasing 93 days cheaper than buying 365 days on average?

This illustrates the reverse anchoring mentioned earlier.

According to conventional logic, buying a year should yield a lower average daily price. However, if users calculate carefully and discover that buying 93 days is cheaper, they may be more inclined to purchase the 93-day tipping option (the last sentence here is a rather awkward description, yes, it refers to purchasing a tip).

The above content may be seen as over-analysis. However, if Kimi Chat’s official team genuinely designed this payment model with such thinking, it can only be said that their product design capabilities are truly impressive.