DeepSeek’s claim to fame is massively shrinking the per-user memory footprint, enabling many more concurrent users in your fixed hardware footprint. Having more concurrent users leads to much higher overall throughput on that same hardware. That’s why DS is so cheap to host.
See table 2 in this article to see the tradeoffs of overall throughput vs per user latency as an example.
Deepseek would let you host higher concurrency than GLM.
Presumably that compression doesn’t come for free on the quality side. (This is totally speculative on my part. I haven’t personally used DS, and I don’t particularly feel benchmarks correlate well to my own experiences)
If you’ve got a small enough user count, perhaps there are tradeoffs in DeepSeek that are not buying you anything.
I'm not sure I gave Deepseek-v4.1-Flash a fair shot, but in my limited testing, it spends 20 minutes thinking about relatively basic prompts, so even if it's cheaper that GLM, the time saved by GLM-5.3's much more reasonable thinking time/tokens means you end up ahead if you value your own time.
Having used both, they really are comparable for most things, which leaves the cheaper Deepseek a better choice for most folks. The one area that I've found GLM to be a clear winner is in offensive Cyber.
If you run a red team GLM is hard to beat, and most of the providers who want to serve an abliterated model cost significantly more so it makes more sense to spend money on hardware.
“some fines” could be a very big deal. Sanctions laws have teeth.
How much of Nvidia’s revenue came from infringing sales to china? If so, that revenue is probably going to disappear as now they will have to really be serious about export controls or face much greater pain. Some estimates I’ve seen is 30% of their revenue perhaps disappearing overnight!
In this case, I doubt it would ever happen. They would be making so much money off this it would form its own little cargo cult in the government from people finding ways to siphon it off. Plus apparently we need AI in everything ever immediately. Would they really take down the company doing that? No way.
Sure, but then that would mean they can now take those chips and sell them to Musk, Zuck, or Dario at inflated prices instead.
And I'm pretty sure any CEO with at least a bachelors degree knows about the De beers diamonds story, so, I really find it hard to believe they will struggle with the concept of maintaining their artifical scarcity/demand.
If 30% of their chips were going to China and now they have to stop that, it means they will have an excess supply of chips and the prices would go down instead of up.
That was the PS3 SPU. It was very fast for its time, but only for the small subset of code that could work within its constraints, and viciously difficult to program for.
See table 2 in this article to see the tradeoffs of overall throughput vs per user latency as an example.
Deepseek would let you host higher concurrency than GLM.
Presumably that compression doesn’t come for free on the quality side. (This is totally speculative on my part. I haven’t personally used DS, and I don’t particularly feel benchmarks correlate well to my own experiences)
If you’ve got a small enough user count, perhaps there are tradeoffs in DeepSeek that are not buying you anything.
reply