Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GLM-5.3-Flash

z.ai

Post reply on HN
Print view View original

511–520 of 605 posts

Re: GLM-5.3-Flash

#511

Post by skohan »

kristianp wrote: post #445
skohan wrote: post #228

Earlier quoted context omitted.

I've had the exact opposite experience. I've been using 3.8 for my daily driver since last week, and I've gradually been giving it more and more complex tasks as it continues to deliver high quality results. Now I am basically handing off large complex features, and 3.8 is doing the planning, task breakdown, implementation and review with just a few notes from my side. The tradeoff is time (especially on RDMA4 hardwa…

Both of you should mention what quant you're using. And as another comment said, what tasks you're doing, i.e. coding, classification, summarizing etc.

I'm using unsloth dynamic Q4_K_XL.

My use-case is coding, currently working on a project with a Rust backend and TS/React/Vite frontend, with probably tens of thousands of lines of code total (including tests).

Re: GLM-5.3-Flash

#512

Post by Alifatisk »

gloomyday wrote: post #509
pkulak wrote: post #503

Earlier quoted context omitted.

What are you guys doing where cost is such a concern? I have a $20 codex subscription and I was able to use it to build a bespoke scheduling website for an acquaintance over three days without even going halfway through my quota. On Sol xhigh. I love hearing about new models, but every time I just don’t know why I should use something worse. I tried some random model on fireworks a week ago, and it immediately went i…

Less powerful models are already extremely capable, so going for the best model is just like buying the most expensive hammer in the shop instead of the functional and well-priced one. Your experience is not representative of their usefulness.

I bet Luna (xhigh) could’ve completed the same task.

Re: GLM-5.3-Flash

#513

Post by johntarter »

kzrdude wrote: post #224
Alifatisk wrote: post #213

Earlier quoted context omitted.

And don't forget the coolest part, DeepSeek, Qwen, Z.ai and Moonshot have almost caught up while being open about their research and their model weights. We can mostly speculate about OAI and Anthropic models, nothing else, how fun huh?

Exactly, DeepSeek, Qwen etc are catching the attention because they put out their tech docs and papers, so we can read about how the models work and what they think their innovation was this time.

Do they publish their distillation strategies on the private frontier models? Just curious.

Re: GLM-5.3-Flash

#514

Post by petesergeant »

nullbio wrote: post #505

Despite what any benchmarks tell you, I'm actually finding GLM-5.3 max to be better than Sol and Fable. Finally bit the bullet and installed OpenCode and OpenRouter and have been experimenting with other models. The labs are clearly benchmaxxing a bit to maintain perceptions. But I don't think they're in the lead anymore in terms of their public offering - although I'm sure what they have behind closed doors is far b…

I use all three every day, and I am absolutely not finding 5.3 to be better. Competent and in the same league as, sure, but not better.

Re: GLM-5.3-Flash

#515

Post by Balinares »

Implicated wrote: post #190
metadat wrote: post #168

Earlier quoted context omitted.

Qwen 3.8 27B is around Opus 4.8 level of capability on the Agentic Intelligence Index (52 vs 57). In my testing the locally hosted Qwen is good enough that looking at a given piece of work output I couldn't tell you which model was behind it. https://artificialanalysis.ai/models/qwen3-8-27b?models=gpt-...

As a counter to that - I've tried various flavors/quants/full weights and Qwen 3.8 27B has been entirely useless at anything non-trivial. Sure - it can do some boilerplate work (though, even armed with a well written spec and working within a very well known framework it went off the rails and did things in a way that were... um... questionable at best) but I don't see it as anything more than a personal assistant st…

Interestingly, I've found the output of Qwen 3.8 27B to be competitive with Opus (4.6-ish anyway, not quite 4.8), but the experience is very different.

Where Opus has seen it before and knows how to do it, Qwen knows how to work it out. It turns out it's surprisingly capable at working things out. The obvious drawback is that it takes tokens and time.

All the same, getting to run something this capable locally is momentous, and suggests to me that streaming tokens from colossal data centers might not be the long term path forward.

Re: GLM-5.3-Flash

#516

Post by rawoke083600 »

DetroitThrow wrote: post #225
metadat wrote: post #168

Earlier quoted context omitted.

Qwen 3.8 27B is around Opus 4.8 level of capability on the Agentic Intelligence Index (52 vs 57). In my testing the locally hosted Qwen is good enough that looking at a given piece of work output I couldn't tell you which model was behind it. https://artificialanalysis.ai/models/qwen3-8-27b?models=gpt-...

Lately I've been throwing tasks at Qwen and a frontier or recently-frontier model (as well as Kimi, GLM, etc) and the smaller parameter models are not really comparable to Opus when it comes to making intelligent decisions about greyer areas of good software architecture. Amazing results for open weight and that size, but a really long way off, and I'm extremely skeptical of benchmarks that show these smaller models…

I got so much better experience LLM-Chunking(think RAG) with qwen-38 27B ONCE i move the thinking effort to HIGH vs XHIGH (i think is the default on Open Router).

Re: GLM-5.3-Flash

#517

Post by Balinares »

cdurth wrote: post #426
skohan wrote: post #228

Earlier quoted context omitted.

I've had the exact opposite experience. I've been using 3.8 for my daily driver since last week, and I've gradually been giving it more and more complex tasks as it continues to deliver high quality results. Now I am basically handing off large complex features, and 3.8 is doing the planning, task breakdown, implementation and review with just a few notes from my side. The tradeoff is time (especially on RDMA4 hardwa…

I can't get 3.8 to exit thinking loops. It will just think and think and think on the most trivial topics. I wanted it to port a speed test powershell script to c#. Claude opus 5 completes it under 60 seconds. I let 3.8 churn about 6 different times for 30+ minutes and it never wrote a single line of code to disk. It wrote lots of lines in thinking. unsloth/Qwen3.8-27B-GGUF UD-Q3_K_XL DSH (pi) Any tips?

I've found it tends toward long thinking loops even for simple tasks (and any quantization seems to increase their length), but those do exit eventually, unlike with Qwen 3.6.

I use the Unsloth UD_Q2_K_XL GGUF with default parameters, along with that custom template linked elsewhere in the thread, and no K/V cache quantization.

Re: GLM-5.3-Flash

#518

Post by ma2kx »

Holy shit, is this model really that bad???

Just asked it a question via the custom opencode go endpoint routed over cloudflare ai gateway doesnt show me the correct models. My fault was that I set "https://opencode.ai/zen/go" as endpoint and tried my-gateway.com/custom-ocgo/v1/models, turns out I had to add /v1 to the opencode url and leave it on my -gateway.com.

But first it told me that opencode is not on the compat endpoint and than it recommendet "Option 3: Nutzung seit Curse die Integration des Providers класный" . Not to mention some complete gibberish like "Falls Opencode.ai in Cloudflare Eing? Wenn ja, wähle den eingebauten provider o.ä. Weil der Gateway dann ein eigenes Modell-List gibt." or "Schlage die Modellnamen in einer Liste ab (z.B. lokal fester Key)"

Is this just cloudflare or is it really that bad? I mean thats not even the level of LLama2 7b ...

Re: GLM-5.3-Flash

#519

Post by Mashimo »

ma2kx wrote: post #518

Holy shit, is this model really that bad??? Just asked it a question via the custom opencode go endpoint routed over cloudflare ai gateway doesnt show me the correct models. My fault was that I set " https://opencode.ai/zen/go " as endpoint and tried my-gateway.com/custom-ocgo/v1/models, turns out I had to add /v1 to the opencode url and leave it on my -gateway.com. But first it told me that opencode is not on the co…

> I mean thats not even the level of LLama2 7b ...

I think you answered your own question. Must be something wrong.

Post reply on HN