The "affordable luxury" framing is the tell. This is where a Chinese price war stops racing to the bottom and someone climbs back up the ladder. I've watched it in coffee, tea and EVs: commoditise to grab the base, then reposition on "value" and charge for it. Kimi pricing to the domestic ceiling while matching frontier performance is that move, aimed at tokens instead of lattes. Whether enterprises pay for stability the way consumers pay for a brand is the real question. My hunch: some will, and that's how this market tiers rather than converging to zero.
Is the “luxury” part mostly about capability per dollar, or about the feeling that you can finally use a strong model without mentally metering every prompt? That second bit feels underrated to me.
Kimi 3 pricing is not getting the attention it deserves. Kudos for recognizing it. To date, revenue has accrued to Frontier models. Deepseek provides a great example. It was launched roughly at the same time as Sonnet. Since then Anthropic has scaled to $80 billion and Deepseek around $1 billion. Kimi 3 is a very good model, but then so was Deepseek V3. I don’t think Kimi 3 will change the landscape. Can it support the current pricing structure? Time will tell, but its the right question.
The "affordable luxury" framing is the tell. This is where a Chinese price war stops racing to the bottom and someone climbs back up the ladder. I've watched it in coffee, tea and EVs: commoditise to grab the base, then reposition on "value" and charge for it. Kimi pricing to the domestic ceiling while matching frontier performance is that move, aimed at tokens instead of lattes. Whether enterprises pay for stability the way consumers pay for a brand is the real question. My hunch: some will, and that's how this market tiers rather than converging to zero.
Fascinating to see how fast these open models are catching up in scale, great chart to visualize the trend.
Is the “luxury” part mostly about capability per dollar, or about the feeling that you can finally use a strong model without mentally metering every prompt? That second bit feels underrated to me.
Kimi 3 pricing is not getting the attention it deserves. Kudos for recognizing it. To date, revenue has accrued to Frontier models. Deepseek provides a great example. It was launched roughly at the same time as Sonnet. Since then Anthropic has scaled to $80 billion and Deepseek around $1 billion. Kimi 3 is a very good model, but then so was Deepseek V3. I don’t think Kimi 3 will change the landscape. Can it support the current pricing structure? Time will tell, but its the right question.