GLM-5.2 is the open-source Claude moment.
The demand we’re seeing at Databricks is astonishing. The world is going to see massive adoption of oss LLMs.
Also, more companies will shift toward post-training their own models on top of oss models and owning the weights.
The demand we’re seeing at Databricks is astonishing. The world is going to see massive adoption of oss LLMs.
Also, more companies will shift toward post-training their own models on top of oss models and owning the weights.
Jun 28, 2026 · 5:03 PM UTC
128
328
3,659
546,051
@Yuchenj_UW Sometimes with every disappointment comes an opportunity. US restrictions have indirectly propelled oss use and adoption. We are witnessing another big change in AI landscape
1
0
85
21,153
@Onwuta_Kelvin as long as the worst case doesn’t happen
best case:
OSS surpasses Mythos, and the gov stops banning GPT-5.6/Fable.
worst case:
OSS surpasses Mythos, and then decides to stop being open source.
4
1
46
28,217
@Yuchenj_UW the GLM 5.2 in databricks is what quantization and what speed currently?
1
0
11
7,934
@aaryan_kakad Performance always vary depending on the workload. We keep working on optimizing it for our customers’ workloads, which can differ case by case.
4
0
12
7,268
@Yuchenj_UW Agreed 100%. Damn, you're speaking my language. Didn't know that Databricks offers these models?
1
0
2
275
@Yuchenj_UW owning the model sounds free until you're the one getting paged at 2am. great for Databricks, less great for your eng team😂
0
0
1
567
@Yuchenj_UW Post-training is the new fine-tuning. What's your go-to stack for efficient weight updates on those OSS models?
0
0
1
199
@Yuchenj_UW This is very different compared to when I was at databricks 😂. Things change quickly
0
0
1
211
@Yuchenj_UW the honest answer is that oss llms are a game-changer because they're the first to make large-scale model training even remotely feasible outside of huge cloud budgets
0
0
1
479
@Yuchenj_UW I utilize a couple OSS models on Cloudflare and it’s incredible how affordable they are. Pennies on the dollar.
0
0
0
115
@Yuchenj_UW Why does anyone use a proprietary closed produce like databricks?
2
1
8
2,868
@Yuchenj_UW To think this is just a 744B parameter model, imagine what will happen when 1.6T parameter Deepseek-4-pro launches in future Ds-4.1 & Ds-4.2 versions with improved performance and maintain it current cheap inference pricing 😄
0
0
6
603
@Yuchenj_UWresting point on GLM-5.2 being such a big moment for open-source LLMs.
We’re seeing the exact same trend — tons of teams are investing in post-training open model weights to own their custom stacks.
Running a mix of self-tuned open models alongside closed ones like Claude does get messy though, switching between different endpoints and billing systems all the time.
We built ApiMax to smooth that workflow out: one single access point for all open and closed global models, unified billing and global stable acceleration, making it easy to route work between premium models and affordable open alternatives as you need.
We’re seeing the exact same trend — tons of teams are investing in post-training open model weights to own their custom stacks.
Running a mix of self-tuned open models alongside closed ones like Claude does get messy though, switching between different endpoints and billing systems all the time.
We built ApiMax to smooth that workflow out: one single access point for all open and closed global models, unified billing and global stable acceleration, making it easy to route work between premium models and affordable open alternatives as you need.
0
1
2
105
@Yuchenj_UW I hope next version of GLM can do more optimization on the agentic task
0
0
4
2,921
@Yuchenj_UWtmiyor ama GLM-5.2'nin açık kaynaklı Claude anı olması, aslında büyük teknoloji şirketleri için korkunç bir kabusun başlangıcı. yıllarca süper bilgisayarlarla interneti ezberletmeye çalıştılar, milyarlar harcadılar. şimdi 15 milyon parametreli, tek gpu'da çalışan ufacık bir model fizik kurallarını öğreniyor.
bu resmen bir "dejavu" anı: tıpkı 2000'lerin başında linux'un microsoft'u terletmesi gibi. sadece bu sefer, "linux" makine öğreniminin ta kendisi. yakında "bizim de bir açık kaynak stratejimiz var" diyen ceo'ları görürsek şaşırmayalım. zaten birçoğu multi-llm stratejisine geçmişti bile.
bu resmen bir "dejavu" anı: tıpkı 2000'lerin başında linux'un microsoft'u terletmesi gibi. sadece bu sefer, "linux" makine öğreniminin ta kendisi. yakında "bizim de bir açık kaynak stratejimiz var" diyen ceo'ları görürsek şaşırmayalım. zaten birçoğu multi-llm stratejisine geçmişti bile.
0
0
2
445
@Yuchenj_UW I think the project @Teknium is working on with Mixture of Models is going to change things even more.
The preliminary stats look very promising! I imagine pairing deepseek v4 pro with Mimo and see what happens!
The preliminary stats look very promising! I imagine pairing deepseek v4 pro with Mimo and see what happens!
1
0
3
528
@Yuchenj_UW Everyone saying glm 5.2 is so great is clearly not using it to do anything important or advanced. The tooling around the AI models is critical.. seriously critical and they suck balls without it
1
0
3
795
@Yuchenj_UW Curious what demand are you referring to? Are you reselling LLMs at Databricks?
0
0
3
1,561
@Yuchenj_UW Databricks is the US database company right? Didn’t realize they are into hosting LLMs
0
0
2
233
@Yuchenj_UW Do you think post training will be more than a lora? I want to believe in fine tuning models.
0
0
2
1,710
@Yuchenj_UW the llama moment but for the frontier
every enterprise that's been waiting for open source to catch up just got their answer
every enterprise that's been waiting for open source to catch up just got their answer
0
0
2
415
@Yuchenj_UW The Ai race is crazy. In 2022, with the first GPT, who actually thought it would only be 4 years until it’s treated like a weapon?
0
0
2
1,302
@Yuchenj_UW @Yuchenj_UW @yuchenj_UW curious to see how smaller teams will handle post-training. feels like it could level the playing field or overwhelm, depending on resources. what's your take?
0
0
1
6
@Yuchenj_UW We’ll see. Not a sustainable approach. At some point, you can’t carry those weights forward and need to restart.
0
0
1
643
@Yuchenj_UW Is it pay per token and is it available in UK regions ?
0
0
1
339