Booz Allen is cute, but if china can train K3 on a fraction of the US compute availability, yet it benchmarks almost equivalent to Fable for a third of the cost, it’s game over for US labs in the long run.
Believe me, I wish this wasn’t the case, but open up the hardware on the device you are reading this on and tell me the majority of tech inside wasn’t made in China….
Manufacturing was lost a long time ago, this is really just another way
> if china can train K3 on a fraction of the US compute availability, yet it benchmarks almost equivalent to Fable for a third of the cost, it’s game over for US labs in the long run.
I agree. However, as of yet, most/all leading PRC models are distilled from US models. I've personally observed Deepseek, GLM, and Kimi all respond that they are Claude when asked, and the networks of tens of thousands of proxy accounts that we've found show that it's happening on a large scale.
But - if the PRC labs actually train up the domain expertise to train those models from scratch, which they are in the process of doing - then the US is cooked. They're not there yet, but it's probably only a matter of years.
Playing devils advocate, does from scratch really matter if all frontier labs are training off each other anyways? Practically speaking, businesses/consumers just care about lowest inference cost for maximizing intelligence anyways (not to mention Anthropic/OpenAI forcing KYC/litigation barrier trash for access to any cybersecurity/IT capabilities) that I literally just cancelled Claude today.
Yeah it’s sad the CCP has my information. But it’s either them or the feds, and at least Chinese models actually work for cybersecurity tasks, not to mention aren’t stupid expensive
Artificialanalysis.ai rn on opus 5 is a joke. The main intelligence benchmarks it is like 1% better than Fable, but the cost difference between that and K3 is so funny lol. It’s the same with cars— you don’t have to do it from scratch, as long as you can do it cheaper and with the same quality, hence Toyota/Honda taking over
Yeah it’s sad that American models are censored more than Chinese ones lol (outside of asking them about the CCP) but it’s where we are at I guess :(
Anthropic wants it that way so they can’t be sued… the correct answer is don’t, switch to Kimi K3 for much more usage, the same quality of model, and no hidden practices :/
As a verbal processor, I hate this. I use whisper v3 large all the time from voice memo dumps that are AI grammar corrected and this would flag when I copy paste it in :/
This would run into the same deal as VINs in the real world being tied to licenses, or serial numbers to guns. But the car equivalent of VMs/open hardware/custom firmware (imagine a $7 pi zero flashed with lineageOS “overage phone”) then becomes equivalent to a gun without a serial number, and suddenly open source/hardware people are felons and there is insane amount of control on hardware and software like it’s 1982.
This assumes that the government would be able to verify independently a phone serial number so that people’s IDs aren’t leaked. If not, then you’re back to the same thing as before since “drivers licenses” are stored by sites and shared around with advertisers
At some point I should post it, I got deepseek to fully reverse engineer the d2 tala engine binary (the sota for diagrams to code with ER ones), and I’m pretty close to an open source looking Go build with 1:1 parity
Once that’s done I guess I can just clean room design it to wasm and post it on npm lol
I’ve been doing this too, it’s a cheat code! 1/100th of the price of Claude/openai prices for 95% of the quality. Site is platform.deepseek.com for that. No image generation, just text, but if you use it right it works great
Believe me, I wish this wasn’t the case, but open up the hardware on the device you are reading this on and tell me the majority of tech inside wasn’t made in China….
Manufacturing was lost a long time ago, this is really just another way
reply