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

Qwen3.6-Plus: Towards real world agents

qwen.ai

Post reply on HN
Print view View original

191–200 of 235 posts

Re: Qwen3.6-Plus: Towards real world agents

#191

Post by teruakohatu »

manc_lad wrote: post #103
simonw wrote: post #66

Pretty solid Pelican: https://gist.github.com/simonw/ca081b679734bc0e5997a43d29fad... I used the https://modelstudio.alibabacloud.com/ API to generate that one, which required signing up for an account and attaching PayPal billing - but it looks like OpenRouter are offering it for free right now so I could have used that: https://openrouter.ai/qwen/qwen3.6-plus:free

they're going to start training a pelican riding a bike specifically on these models soon. it's the key global benchmark!

They will have a long time ago. By now Simon's meme will be well represented in training sets.

Re: Qwen3.6-Plus: Towards real world agents

#193

Post by kadoban »

drob518 wrote: post #131
zozbot234 wrote: post #35

Earlier quoted context omitted.

> not an open weight model like they’ve become known for. Right, they state that they'll release "smaller" variants openly at some point, with few details as to what that means. Will there be a ~300B variant as with Qwen 3.5? The blog post doesn't say.

As a publicity stunt, releasing a 300B open model is pretty smart. You can talk about its strong performance and it being “open” and “available,” but it’s so large that most people can’t use it themselves and might try out the cloud-based offering.

I'm running qwen 3.5 397b on very standard hardware. Just use the unsloth quants, they're great. I get like 20t/s or something.

It's super not a publicity stunt, qwen 3.5 is the base of the best local models out there IMO.

Re: Qwen3.6-Plus: Towards real world agents

#194

Post by breisa »

Aerroon wrote: post #185
edg5000 wrote: post #171

Do they have an API where you can control the chat template or at least just put everything in the system prompt? This way you can control everything including the tool calling syntax. Even if you use the trained tool syntax, it allows you to control the tool system prompt which you may want to tweak. With DeepSeek this is all possible. An undocumented feature, great for harness builders. Anybody got info on Qwen reg…

Can't you do that on OpenRouter? You can set a system prompt there. Is that insufficient for what you had in mind?

They are talking about the chat template and not the system prompt. With current gen models, the system prompt is only part of a larger pretext that is passed to the model at the start of the "chat". The models are trained on a specific chat template with things like tool lists, reasoning budget, special feature flags and the "system prompt" formatted in a certain template.

Re: Qwen3.6-Plus: Towards real world agents

#195

Post by jona-f »

Aurornis wrote: post #11

This is their hosted-only model, not an open weight model like they’ve become known for. They got a lot of good publicity for their open weight model releases, which was the goal. The hard part is pivoting from an open weight provider to being considered as a competitor to Claude and ChatGPT. Initial reactions are mostly anger from everyone who didn’t realize that the play along was to give away the smaller models as…

> Most users of cheap API tokens are not loyal to any brand

In the exploration phase, yes. But once your setup settles down you likely want to stay on the same model for stable operation.

Re: Qwen3.6-Plus: Towards real world agents

#196

Post by kiviuq »

andsoitis wrote: post #129
Filligree wrote: post #117

Earlier quoted context omitted.

China has never threatened war against my country; America has. Between the two, it’s clearly safer to lean towards the Chinese options if EU ones aren’t available.

That’s incredibly naïve.

Nothing but Love for China.

Re: Qwen3.6-Plus: Towards real world agents

#197

Post by abc123abc123 »

mikrotikker wrote: post #156
cmrdporcupine wrote: post #108

Earlier quoted context omitted.

Whereas I as a Canadian am absolutely eager to see a serious competitor from a rival to the US because sending money south to Anthropic and OpenAI who think it's ok to spy on (or worse) their non-American customers, and are headquartered in a country that is trying to crush my country's economy, interfere in our domestic politics, and put us out of work and making threats on political allies. I'd prefer them to be op…

You're giving up the rest of your country to a geopolitical rival from a separate region, in a separate hemisphere with smiling expansionist goals, even allowing armed Chinese security to protect Chinese installations in country. So why not give the rest of your country to China. It will help them get a good flank on the USA such that even when that temporarily embarrassed country gets a leader you, and the rest of t…

No, a rational decision based on a crazy man in the US. The US needs to learn, that if it threatens its traditional allies, they go to work with china, the main competitor of the US. If the US wants it allies back, the tariffs have to go, and the childlike rhetoric and threats as well. If not, china _deserves_ the business of the US former allies.

Re: Qwen3.6-Plus: Towards real world agents

#198

Post by imtringued »

senordevnyc wrote: post #71
jjice wrote: post #34

Earlier quoted context omitted.

OpenRouter usage is likely skewed towards LLMs that are more niche and/or self-hostable by solid hardware that's available, but most consumers don't have on hand. I can imagine Anthropic and OpenAI LLMs often get called directly from their APIs instead. At least from my experience and friends of mine, we use OpenRouter for cases where we want to use smaller LLMs like Qwen, but when I've used ChatGPT and Claude, I use…

Same, and my little SaaS is pushing more than 0.1% of the TOTAL volume of tokens on OpenRouter, so the reality is they’re TINY.

0.1% of OpenRouter is around 400 billion tokens per month or around $400k per month at a cost of $1 per 1 million input tokens, not counting output.

I think it's pretty disingenuous to call your SaaS little when it is projected to spend at least 5 million USD just on tokens and this is a low end estimate.

Re: Qwen3.6-Plus: Towards real world agents

#199

Post by dryarzeg »

justinclift wrote: post #127
wg0 wrote: post #67

It hallucinates a lot more then Sonnet or even MiniMax M2.5. Especially in tool calls, it would end up duplicating the content in code files and then realising later and getting stuck in a loop.

> It hallucinates a lot more then Sonnet or even MiniMax M2.5. Ugh, that's not good. I evaluated Kimi K2 a while back for some text understanding -> summarisation tasks, and of the 100 tasks it hallucinated about 30% of the output. :( :( :(

> I evaluated Kimi K2 a while back

I guess that it was Kimi K2-Instruct, the first model (or it's fine-tune) in the lineup of Kimi-K2 models. And I remember trying it just for the sake of curiosity, and... except for the almost total absence of the sycophancy and "sugar syrup" in it's outputs, it was not very good at the time. Right now though, if you're still interested in this model family, you could look at Kimi-K2.5 which is way better.

That said, it's still not perfect, and to be honest, looking where things are going with LLMs right now I prefer the use of my own brain (local private inference with power consumption of ~20-25W, having a capability for continuous learning and performing real-world tasks) to the use of any "AI" model (including proprietary models such as Claude 4.6 Opus, Gemini 3.1 Pro and others).

: )

Re: Qwen3.6-Plus: Towards real world agents

#200

Post by SyneRyder »

nwienert wrote: post #112
Aurornis wrote: post #11

This is their hosted-only model, not an open weight model like they’ve become known for. They got a lot of good publicity for their open weight model releases, which was the goal. The hard part is pivoting from an open weight provider to being considered as a competitor to Claude and ChatGPT. Initial reactions are mostly anger from everyone who didn’t realize that the play along was to give away the smaller models as…

4.5 is better than 4.6 though in practice. 4.6 was purely a cost savings change with enough benchmark gamification to look better.

I've found Opus 4.6 to be smarter than 4.5, at least in some ways. There's a bug I'd been trying to solve for a decade (and so had other humans) and I've been giving it to each model to try and solve, including in interactive sessions. Each model got closer, but none of them actually solved it, until Opus 4.6 got it on the first go (I probably used Ultrathink). This was before the 1M context was available.

I'd agree that 4.6 and 4.5 are different, but I don't think it's correct that 4.6 is just reduced and benchmaxxed. It genuinely solved problems for me that no other model has been able to.

I think I'd like to have seen the 4.6 benchmarks also included against Qwen.

Post reply on HN