At this point, they are beginning to approach the Tesla Roadster case. The flying one.-

to me , glm flash and deep seek is that chinese 3 wheel tuk-tuk or e-bike runs on electricity. It make perfect sense to run to supermarket, and it's a backbone of Chinese economy, but no way in my life for a chance to save $100/mo you will risk your business with bugs, and bad/worse architecture decisions to save that $100.

again, they can be great for narrow use cases, and save you a fortune.

PRC models are more like ubiquitous class8 trucks that all economies can use as backbone. SOTA AI is like sending a courier on a first class ticket, only good for niche high ROI scenarios. No one is haggling over $100 per month, but $1000s per head per month on enterprise SOTA vs PRC models. For subsidized $100 per month plans SOTA vendors goes broke anyways... at premium SOTA API most users go broke. Pareto frontier PRC models are being routed to take over all the general use cases, seems like essentially everyone outside of niche coding is scaling back SOTA AI use because it's not worth the $$$. People figured out you don't need next day airfreight everywhere.

Deepseek Flash is not limited to "narrow use cases" by any means. If we went into AI winter tomorrow I'd be surprised if Flash V4.1 wouldn't be a perfectly functional drop-in replacement for majority of current token spending.

In some ways it's better/faster at routine CLI/sysadmin stuff I do daily than Sol 6.1. It's surprisingly "clever" when hitting obstacles and particularly persistent unlike say Luna. It really is a remarkably capable agentic tuned model even before accounting for size.

   few use cases really need it.

Unfortunately, I came to the conclusion that higher intelligence is extremely beneficial even for 'dumb tasks'. So I resorted to switching to Opus for pretty much everything.

when you're on a $100/mo plan, this is absolutely true. when you're paying API rates to deliver a product to people at scale and the difference between $1 and $0.03 per response is the difference between a feature being viable or not, you realize there's lots of things that haiku is fine for.