3 ms·
GLM-5.2: Frontier Intelligence, Open Weights
- Bucko1 4mo agoI have already tested this and it is impressive.
- dbbk 4mo agoIf I have a fully maxed out MacBook Pro, would it make sense to just switch from Opus 4.8 to this? I've never tried running local models for coding...
- entrope 4mo agoHuggingFace says this model has 753B parameters, which will need a lot more RAM than a maxed-out MacBook Pro. With 40B active parameters, running from SSD would need patience.
- api 4mo agoI’ve wondered for a while if anyone is working on very wide channel parallel (kind of like RAID 0) SSD for this purpose. Couple that with a tensor processor and that would be interesting.
- petu 4mo agoThere's talks about HBF. F for Flash -- HBM packaging and bus width, but using NAND memory. e.g. https://www.sandisk.com/company/newsroom/press-releases/2026/2026-02-25-sandisk-and-sk-hynix-begin-global-standardization-of-next-generation-memory-solution-high-bandwidth-flash-hbf https://www.sandisk.com/company/newsroom/press-releases/2026...
- _aavaa_ 4mo agoFor an fp4 quantization it should fit with room to spare for KVCache
- hadik 4mo agoBlog post with more information including benchmarks: https://z.ai/blog/glm-5.2 https://z.ai/blog/glm-5.2
- osti 4mo agoGiven that DeepSwe is one of the very few coding benchmarks worth taking a look at, this achieves rather excellent result at it (not far from opus 4.8). From looking at the results and my own impression of 5.1 and other models, I think this is the best Chinese coding model by some non-insignificant margin.
- LaurensBER 4mo agoI've been very pleased with it's performance over the last few days. It's definitely not near Opus 4.8 level but it's very impressive nonetheless and it does do design extremely well.
- ebbi 4mo ago> it does do design extremely well Better than Opus?
- osti 4mo agoI don't know what people mean when they say design lol, is it for frontends?
- ebbi 4mo agoYeah, that's what I mean anyway. Each model has certain design tropes it repeats everywhere, and some of them are very old-school or not really UI best practice. And then the more ambitious cases where you ask for a feature without being prescriptive with UI needs, the end result is sometimes atrocious with weird font use, colours, etc.
- fcanesin 4mo agoZhipu AI is founded by a superstar Tsinghua professor, did an IPO in January (Hong Kong stock exchange) hired half it's past research lab and it's stock is >10x since. This is not a "just distill Claude" thing.
- saxenaabhi 4mo agoIPO within year of founding? It it normal for startup in China?
- ChrisArchitect 4mo agonon-x blog post: https://z.ai/blog/glm-5.2 https://z.ai/blog/glm-5.2 And previously: https://news.ycombinator.com/item?id=48518684 https://news.ycombinator.com/item?id=48518684
- gnabgib 4mo agoDiscussion (765 points, 3 days ago, 489 comments) https://news.ycombinator.com/item?id=48518684 https://news.ycombinator.com/item?id=48518684
- coolspot 4mo agoLooks like their code subscription is sold out. Clicking on the "Subscribe" button just does nothing. Pay-as-you-go API works well.