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

GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

magazine.sebastianraschka.com

Post reply on HN
Print view View original

51–60 of 99 posts

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#51

Post by xfalcox »

cranberryturkey wrote: post #31
7moritz7 wrote: post #3

Qwen3 is substantially better in my local testing. As in, adheres to the prompt better (pretty much exactly for the 32B parameter variant, very impressive) and is more organic sounding. In simplebench gpt-oss (120 bn) flopped hard so it doesn't appear particularly good at logical puzzles either. So presumably, this comes down to... - training technique or data - dimension - lower number of large experts vs higher num…

qwen3 is slow though. i used it. it worked, but it was slow and lacking features.

Qwen 3 is not slow by any metrics.

Which model, inference software and hardware are you running it on?

The 30BA3B variant flies on any GPU.

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#52

Post by omneity »

7moritz7 wrote: post #3

Qwen3 is substantially better in my local testing. As in, adheres to the prompt better (pretty much exactly for the 32B parameter variant, very impressive) and is more organic sounding. In simplebench gpt-oss (120 bn) flopped hard so it doesn't appear particularly good at logical puzzles either. So presumably, this comes down to... - training technique or data - dimension - lower number of large experts vs higher num…

Qwen3 32B is a dense model, it uses all its parameters all the time. GPT OSS 20B is a sparse MoE model. This means it only uses a fraction (3.6B) at a time. It’s a tradeoff that makes it faster to run than a dense 20B model and much smarter than a 3.6B one.

In practice the fairest comparison would be to a dense ~8B model. Qwen Coder 30B A3B is a good sparse comparison point as well.

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#53

Post by selcuka »

omneity wrote: post #52
7moritz7 wrote: post #3

Qwen3 is substantially better in my local testing. As in, adheres to the prompt better (pretty much exactly for the 32B parameter variant, very impressive) and is more organic sounding. In simplebench gpt-oss (120 bn) flopped hard so it doesn't appear particularly good at logical puzzles either. So presumably, this comes down to... - training technique or data - dimension - lower number of large experts vs higher num…

Qwen3 32B is a dense model, it uses all its parameters all the time. GPT OSS 20B is a sparse MoE model. This means it only uses a fraction (3.6B) at a time. It’s a tradeoff that makes it faster to run than a dense 20B model and much smarter than a 3.6B one. In practice the fairest comparison would be to a dense ~8B model. Qwen Coder 30B A3B is a good sparse comparison point as well.

> GPT OSS 20B is a sparse MoE model. This means it only uses a fraction (3.6B) at a time.

They compared it to GPT OSS 120B, which activates 5.1B parameters per token. Given the size of the model it's more than fair to compare it to Qwen3 32B.

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#54

Post by vidarh »

threeducks wrote: post #35
yunusabd wrote: post #24

Earlier quoted context omitted.

GP asked the model to _create_ a riddle, not solve a given one.

Yes, but the odds of getting GPT-OSS to respond with that riddle are pretty low and it is not necessary to demonstrate whether the LLM can answer the riddle correctly.

They said it provided the answer when it created the riddle. They didn't question itd ability to solve it.

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#55

Post by tough »

indigodaddy wrote: post #33
lvl155 wrote: post #26

Earlier quoted context omitted.

Qwen3 coder 480B is quite good and on par with Sonnet 4. It’s the first time I realized the Chinese models are probably going to eclipse US-based models pretty soon, at least for coding.

Where do you use qwen3 480b from, I'm not even seeing it on Openrouter. EDIT nm, openrouter is just calling it qwen3-coder-- when I click for more info it shows it's Qwen3-Coder-480B-A35B-Instruct. And it's one of their free models. Nice

cerebras code (both sub and api) have it

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#56

Post by faangguyindia »

7moritz7 wrote: post #3

Qwen3 is substantially better in my local testing. As in, adheres to the prompt better (pretty much exactly for the 32B parameter variant, very impressive) and is more organic sounding. In simplebench gpt-oss (120 bn) flopped hard so it doesn't appear particularly good at logical puzzles either. So presumably, this comes down to... - training technique or data - dimension - lower number of large experts vs higher num…

yesterday, i signed up for qwen3-coder-plus. It fails 4/10 "diff" edit format in various code editing tools i use.

Gemini Pro 2.5 with diff fenced edit format, rarely fails. So i don't see this Qwen3 hype unless i am using wrong edit format, can anyone tell me which edit format will work better with Qwen3?

https://aider.chat/docs/more/edit-formats.html

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#57

Post by faangguyindia »

jszymborski wrote: post #4
7moritz7 wrote: post #3

Qwen3 is substantially better in my local testing. As in, adheres to the prompt better (pretty much exactly for the 32B parameter variant, very impressive) and is more organic sounding. In simplebench gpt-oss (120 bn) flopped hard so it doesn't appear particularly good at logical puzzles either. So presumably, this comes down to... - training technique or data - dimension - lower number of large experts vs higher num…

If I had to make a guess, I'd say this has much, much less to do with the architecture and far more to do with the data and training pipeline. Many have speculated that gpt-oss has adopted a Phi-like synthetic-only dataset and focused mostly on gaming metrics, and I've found the evidence so far to be sufficiently compelling.

this is exactly why strongest model gonna lose out to weaker models if the later ones have more data

for example, i was using deep seek webui and getting decent on point answers but it simply does not have latest data.

So, while Deep Seek R1 might be better model than Grok3 or even Grok4, it not having access to "twitter data" basically puts it behind.

Same is case with OpenAI, if OpenAI has access to fast data from github, it can help with bugfixs which claude/gemini2.5 pro can't.

model can be smarter but if it does not have the data to base its inference upon it's useless.

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#58

Post by faangguyindia »

tough wrote: post #55
indigodaddy wrote: post #33

Earlier quoted context omitted.

Where do you use qwen3 480b from, I'm not even seeing it on Openrouter. EDIT nm, openrouter is just calling it qwen3-coder-- when I click for more info it shows it's Qwen3-Coder-480B-A35B-Instruct. And it's one of their free models. Nice

cerebras code (both sub and api) have it

what edit format u use with Qwen? https://aider.chat/docs/more/edit-formats.html

diff is failing me or do you guys use whole?

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#59

Post by SchemaLoad »

cranberryturkey wrote: post #31
7moritz7 wrote: post #3

Qwen3 is substantially better in my local testing. As in, adheres to the prompt better (pretty much exactly for the 32B parameter variant, very impressive) and is more organic sounding. In simplebench gpt-oss (120 bn) flopped hard so it doesn't appear particularly good at logical puzzles either. So presumably, this comes down to... - training technique or data - dimension - lower number of large experts vs higher num…

qwen3 is slow though. i used it. it worked, but it was slow and lacking features.

GPT-OSS is slow too. Gemma3 gives me better results and runs faster.

Re: GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

#60

Post by Mars008 »

selcuka wrote: post #53
omneity wrote: post #52

Earlier quoted context omitted.

Qwen3 32B is a dense model, it uses all its parameters all the time. GPT OSS 20B is a sparse MoE model. This means it only uses a fraction (3.6B) at a time. It’s a tradeoff that makes it faster to run than a dense 20B model and much smarter than a 3.6B one. In practice the fairest comparison would be to a dense ~8B model. Qwen Coder 30B A3B is a good sparse comparison point as well.

> GPT OSS 20B is a sparse MoE model. This means it only uses a fraction (3.6B) at a time. They compared it to GPT OSS 120B, which activates 5.1B parameters per token. Given the size of the model it's more than fair to compare it to Qwen3 32B.

You call it fair? 32 / 5.1 > 6, it's takes 6 times more to compute each token. Put it other way, Qwen3 32B is 6 times slower than GPT OSS 120B.
Post reply on HN