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 advertising, not because they were feeling generous.
Comparing to Opus 4.5 instead of the current 4.6 and other last-gen models is clearly an attempt to deceive, which isn’t winning them any points either.
I think there is a moderately large market for models like this that aren’t quite SOTA level but can be served up much cheaper. I don’t know how successful they’ll be in the race to the bottom in this market niche, though. Most users of cheap API tokens are not loyal to any brand and will change providers overnight each time someone releases a slightly better model.
> 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.
I'm not interested in adopting an inferior closed source weight from a geopolitical rival. The open source weights argument was the one thing China had going and that I was seriously cheering them on for. They could have been our saviors and disrupted the US tech giants - and if it was open, I'd have welcomed it.
Now they show their true colors. They want to train models on our engineering to replace us, while simultaneously giving nothing back? No thanks. I'd rather fund the shitty US hyperscalers. At least that leads to jobs here.
If there's a company willing develop and foster large scale weights in the open, I'll adopt their tooling 100%. It doesn't matter if they're a year behind. Just do it open and build an entire ecosystem on top of it.
The re-AOLization of the internet into thin clients is bullshit, and all it takes is one player to buck the rules to topple the whole house of cards.
> I'm not interested in adopting an inferior closed weights model from a geopolitical rival.
That's a very reasonable stance. It doesn't change the fact that we do have plenty of local models (up to and including Qwen 3.5) that are still quite useful.
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 open weight, but I'd love to sub a decent competitive coding plan from a European or Chinese provider. Right now they're not quite there. If closing it and charging for it brings them closer to competitive, that's ok.
If the US tech and AI industry long term wants customers and a broad market outside of their own domestic base, they need to reconsider who they are bending the knee to, and how they are defining their policies in relation to the Trump administration.
I’m starting to wonder where the most is for any of these models.
Sure they are not cheap to train. But if open weight models continue to be trained and continue to become available on cheaper hardware, how do dedicated AI companies protect their margins?
> Initial reactions are mostly anger from everyone who didn’t realize that the play along was to give away the smaller models as advertising, not because they were feeling generous.
The naivety around this has been staggering quite frankly. All of a sudden, people thinking that meta etc are releasing free models because they believe in open access and distribution of knowledge. No, they just suck comparatively. There is nothing to sell. Using it to recruit and generate attention is the best play for them.
For a brief moment there were a lot of comments about how Chinese tech companies are our saviors in the age of AI because they were releasing their models. It was an edgy contrarian take that was getting a lot of traction, mostly from commenters who were unfamiliar with Alibaba and thought it was the anti-Big-tech
I'm not frustrated or disappointed, we have lots of models from Qwen already. We haven't really lost anything. And plenty of players only release "smaller" models anyway, so it's hardly unprecedented.
I thought Qwen was releasing open-weight because China can't compete with America (because of people's privacy concerns), so the only thing they could do is salt the ground economically with open models, and make sure everybody loses.
Qwen is actually a pretty strong player in the Chinese market. There is an implied "salt the ground" play but it's mostly from hardware makers, who are trying to keep the big AI players honest and also stand to gain if local inference becomes popular.
Opus was released in Feb 2026. Even though it feels like a long 2 months has passed, its' not really clear that they were developing this as a competitor to that product.
There's nothing really strange about not competing directly with the best, but rather showing whom you are as good as.
I don’t know why anyone would do the mental backflips to defend this.
They posted charts with logos for Claude and others. You had to read the fine details to realize they weren’t comparing to the latest offerings from those companies. They were counting on you not noticing.
There’s zero reason to compare to old models unless you’re trying to mislead.
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 those APIs directly.
No. Right now I'm upset that Google has removed (or at least is in the process of removing) the Gemini 2.0 flash model. We use it for some pretty basic functionality because it's cheap and fast and honestly good enough for what we use it for in that part of our app. We're being forced to "upgrade" to models that are at least 2.5 times as expensive, are slower and, while I'm sure they're better for complex tasks, don't do measurably better than 2.0 flash for what we need. Yay. We've stuck with the GCP/Gemini ecosystem up until now, but this is kind of forcing us to consider other LLM providers.
this is one of the reasons im hearing more and more people are using open/locally hosted models. particularly so we dont have to waste time to entirely redo everything when inevitably a company decides to pull the rug out from under us and change or remove something integral to our flow, which over the years we've seen countless times, and seems to be getting more and more common.
products entirely disappearing or significantly changing will be more and more common in the llm arena as things move forward towards companies shutting down, bubbles deflating, brand priorities drastically reshifting, etc...
i think, we're at or at least close to a time to really put some thought into which pieces of your flow could be done entirely with an open/local model and be honest with ourselves on which pieces of our flow truly needs sota or closed models that may entirely disappear or change. in the long run, putting a little bit of thought into this now will save a lot of headache later.
> There isn't, pretty much everyone wants the best of the best.
For direct user interaction or coding problems, perhaps. But as API calls get cheaper, it becomes more realistic to use them for completely automated workflows against data-sets, or as sub-agents called from expensive SOTA models.
For example, in Claude, using Opus as an orchestrator to call Sonnet sub-agents, is a popular usage "hack." That only gets more powerful, as the Sonnet equivalent model gets cheaper. Now you can spawn entire teams of small specialized sub-agents with small context windows but limited scope.
I did create my own MCP with custom agents that combine several tools into a single one. For example, all WebSearch, WebFetch, Context7 exposed as a single "web research" tool, backed by the cheapest model that passes evaluation. The same for a codebase research
Use it with both Claude and Opencode saves a lot of time and tokens.
> But as API calls get cheaper, it becomes more realistic to use them for completely automated workflows against data-sets
Seems like a huge waste of money and electricity for processes that can be implemented as a traditional deterministic program. One would hope that tools would identify recurrent jobs that can be turned into simple scripts.
For example: "Here our dataset that contains customer feedback comment fields; look through them, draw out themes, associations, and look for trends." Solving that with a deterministic program isn't a trivial problem, and it is likely cheaper solved via LLM.
maybe there isnt, but as understanding grows people will understand that having an orchestration agent delegate simple work to lesser agents is significant not only for cost savings, but also for preserving context window space.
That isn't true. In a Codex or Claude Code instance, sure... but those are not the main users of APIs. If you are using LLMs in a service for customers, costs matter.
The market for API tokens is bigger than people like you and I (who also want the best) using then for code.
There are a lot of data science problems that benefit from running the dataset through an LLM, which becomes bottlenecked on per-token costs. For these you take a sample subset and run it against multiple providers and then do a cost versus accuracy tradeoff.
The market for API tokens is not just people using OpenCode and similar tools.
Nope. I get very good results from GLM 5 and 5.1. I’m not working on anything so complex and groundbreaking that I need the best.
Coding is a rung on the ladder of model capability. Frontier models will grow to take on more capabilities, while smaller more focused models start becoming the economical choice for coding
Not really. It depends on the usecase. For private stuff I'm very happy to take what was SOTA a year or 2 ago if I can have it all running in my home and don't have to share any of my data with some sleazy big tech cloud.
The price is a concern too of course. But privacy is a bigger one for me. I absolutely don't trust any of their promises not to use data for training purposes.
OP didn't say about confusing Opus with Qwen but rather people being confused about Qwen3.6-Plus not being available as an "open weight" model available for self hosting.
I understand peoples reactions of Qwen team comparing against Opus 4.5 instead of 4.6. And them comparing against Gemini Pro 3.0 instead of 3.1. But calling it misleading is a bit of stretch in my eyes, people here are acting like we immediately forgot how previous generations performed just because a new version is released.
This field is going in a incredible pace, the providers release a new model every quarter or so. The amount of criticism is a bit overblown in my opinion. The benchmarks still look very good to me. I’ve used GLM-5 (latest is GLM-5.1) and Kimi K2.5, they are decent and gets the job done, so seeing how this model of Qwen performs compared to it is kinda impressive.
Also, why are so many pointing out the fact that this model is not open-weight as if this is their first time doing so. Qwen-3.5-plus, Qwen-3-Max is also closed source. This is not something new.
I think Qwen trying to catch up to the SOTA models is still healthy for us, the consumers. Sure, its sad news that this version is closed-weight, but I won’t downplay their progress.
I think it’s more the principle of deception that upsets people. Imagine if Apple released a new iPhone and publicly compared its specs to some previous gen Android. It’s not in good faith.
They compared their M-series chips to older Intel Macs for a while, likely to target users who were still on Intel chips. If they released a lower cost iPhone and compared it to a previous gen Android I could see the reasoning for it. It's not deception if it's a valid comparison and people just fail to understand what's being compared.
Now, is it mildly deceptive because all of the companies using incredibly confusing naming conventions for their models? Maybe!
Apple continues to compare to prior versions of Apple Silicon. I suspect it is a mix of trying to provide useful, realistic upgrade information and numbers that still sound good for those not paying attention.
I don't think any org doing this is necessarily being deceptive, so long as there's some reasonable basis for the chosen comparable(s).
For example, comparing a new iPhone to a prior Android phone might make sense if the install base is considerably large and Apple is targeting the cohort for user acquisition. (~"These benchmarks are not for you.")
The community will always run the numbers and get the clicks for the benchmarks not filled in by the 1st party. I noticed what appeared to be some movement from Apple in content they've produced to get ahead of this with recent product content.
Why are we so quick to call it deception? Their figure is quite clear. They aren't fiddling with the graph or hiding the labels, they are clearly stating which models it compares against. But I agree on the sentiment that the standard practice should be to bench against the latest SOTA models.
Worth noting that this model, unlike almost all qwen models, is not open-weight, nor is the parameter count exposed. Also odd that it is compared against opus 4.5 even though 4.6 was released like 2 months ago.
"[...] In the coming days, we will also open-source smaller-scale variants, reaffirming our commitment to accessibility and community-driven innovation. [...]"
Qwen3.5-Plus is the largest variant of the open weight Qwen3.5 model, expanded with a 1M context window and fine-tuned on the Qwen-native harness’ specific tools.
In a practical sense, I'm primarily interested in small to medium sized models being open. I think that might be common sentiment.
However, my hope is that there will be at least somewhat competitive big and open models as well, from an ethical/ideological perspective. These things were trained on data that was provided by people without their consent, so they should at least be be publicly accessible or even public domain.
I'll diverge from some of these comments, I don't find it misleading to compare to Opus 4.5.
I can remember how good Opus 4.5 was. If I'm considering using this, it's most informative to me to compare to the model it's closest to that I have familiarity with.
I'm obviously not switching to this if I want the best model. I'm switching if I'm hopeful that the smaller versions are close to it, or if I want to have more options for providers, or for any other reasons unrelated to getting the highest quality responses possible.
Exactly this. If you can get something close to Opus 4.5 for free, that's noteworthy. I may not use it for the most critical pieces of my app, but not everything I do is galaxy-brain coding.
Yes, honestly, Opus 4.6 and GPT 5.4 were mostly not really noticeable improvements over 4.5 and 5.3 respectively. If we were stuck at 4.5 levels but at 1/10th of the price, I'll take it.
I can see reasons, among others that 4.5 was the one established as they were preparing this version. "So long" is merely 2 months ago, and Qwen 3.5 was barely released less than 2 months ago. They were likely already working on finalizing 3.6 before 3.5 official launch, and as 4.6 came out.
In any case, aside Claude fanboyism, having other plays inch closer to similar performance is always useful. Even if they are "6 months behind" as the pace slows down, this guarantees that there's no huge moat and they'll eventually either get to where the SOTA is, or the difference wont be that big.
I'd rather put fewer eggs in 2-3 big player baskets.
3.5-plus was also only available via api. I don’t know what the long term business model for open weights is, I hope there is one, but it seems foolish to assume that companies will be willing to spend millions of dollars of compute on an asset worth zero in perpetuity.
Looking forward to when this gets on Bedrock. I built an app with a niche AI agent and to this point only Sonnet is really good enough for our use case, but its expensive!
I don't know how well it performs, but you can extend Qwen3.5 to 1 million token context using YaRN. Also, Nemotron 3 Super was recently released and scales up to 1 million token context natively.
The agent benchmarks here are interesting but I'd love to see how Qwen3.6-Plus handles long-horizon tasks where it needs to recover from its own mistakes. Most agent evals test the happy path. The hard part is when the model takes a wrong action at step 3 and needs to recognize and backtrack at step 15. Has anyone stress-tested this in a real dev workflow?
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.
I would love to hear from people using both (Claude Code OR Codex) AND (Qwen) and their experience with Qwen models, are they on par, or how far are they?
I switch between Claude Code (Opus/Sonnet) and Qwen (OpenCode, OpenClaw) multiple times throughout the day and Qwen 3.5 is really nice. I do also use KimiK2.5 and GLM5 pretty often too and I'm starting to get a sense that the agent tool is becoming a little more important than the model with these level of models. As long as tool calling and prompt quality is all configured correctly by the provider.
A bit off-topic but I’m on the legacy Lite plan (now discontinued), and it’s more than enough for hobby projects. The main draw is the generous request-based quota (18k requests/month) rather than a token-based one.
This means a 100k token request counts the same as a 100-token one. I’ve made about 8000 requests in the last two weeks, averaging around 80k tokens per request. It feels like they’re subsidizing this just to gather data on agentic workflows.
On the downside, the speed is mediocre (15–30 tg/s for GLM-5), and I’ve seen the model glitch or produce broken output about 10 times out of those 8k requests.
Quite strong results in the benchmarks but why Gemini 3 Pro instead of 3.1? Why only for a few of the benchmarks? Why is OpenAI not there in the coding benchmarks? Why Opus 4.5 and not 4.6? Just jumps out into my eye as a bit strange.
As always, we'll have to try and see how it performs in the real world but the open weight models of Qwen were pretty decent for some tasks so still excited to see what this brings.
I had the exact opposite reaction. I stopped using OpenAI/Google a while ago due to privacy and moved to local Qwen, now I'm considering using Alibaba cloud. You know Google and OpenAI are going to share everything with the US government and Western ad networks. But with Alibaba, who cares if the CCP & Chinese ad networks have a comprehensive profile on me? From a pragmatic perspective it's much better for (outcomes related to) privacy.
so if China has the data good, us has the data bad, got it lol.
us actually has laws around this and they arent sharing very much with thr us gov today. china shares 100% as required by law. and neither care much about "how long do i cook eggs for", but they do care about code generation a lot.
From an espionage perspective your own government is the safest. But from a civil rights perspective your own government is your most immediate threat. China isn't going to arrest me for my opinions on Netanyahu, my own government could
And the US government has repeatedly shown that it is very interested in collecting all the data available, just like China. In China this is simply done in the open while the US has a veneer of protection for citizens. But where the data collection is forbidden by law they either ignore the law or ask another five eyes member to do the spying and share the results. Both are well documented
> China isn't going to arrest me for my opinions on Netanyahu, my own government could
I don't know whether you really believe this or it was an off the cuff remark. China is not going to tell you why they plan to arrest you. China is not a benevolent dictatorship.
> so if China has the data good, us has the data bad
It's not that, it's about relative risk to your own life. Asking questions about "DEI" for example is much more likely to have adverse effects on your life if you ask Grok or an OpenAI chatbot, though still not that likely.
So I guess if it’s your personal data that’s up to you, but if you have private client data and client of your client data and that’s the fundamental reason why you are doing local ai, I can’t imagine moving from qwen local to qwen alibaba after not choosing google/anthropic/openai
As with all arguments equivalent to "I have nothing to hide, so I have nothing to fear," it may be true now, but it may not be true later. The only certainty is that this will not be your call.
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 advertising, not because they were feeling generous.
Comparing to Opus 4.5 instead of the current 4.6 and other last-gen models is clearly an attempt to deceive, which isn’t winning them any points either.
I think there is a moderately large market for models like this that aren’t quite SOTA level but can be served up much cheaper. I don’t know how successful they’ll be in the race to the bottom in this market niche, though. Most users of cheap API tokens are not loyal to any brand and will change providers overnight each time someone releases a slightly better model.
[delayed]
> 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.
I'm not interested in adopting an inferior closed source weight from a geopolitical rival. The open source weights argument was the one thing China had going and that I was seriously cheering them on for. They could have been our saviors and disrupted the US tech giants - and if it was open, I'd have welcomed it.
Now they show their true colors. They want to train models on our engineering to replace us, while simultaneously giving nothing back? No thanks. I'd rather fund the shitty US hyperscalers. At least that leads to jobs here.
If there's a company willing develop and foster large scale weights in the open, I'll adopt their tooling 100%. It doesn't matter if they're a year behind. Just do it open and build an entire ecosystem on top of it.
The re-AOLization of the internet into thin clients is bullshit, and all it takes is one player to buck the rules to topple the whole house of cards.
> I'm not interested in adopting an inferior closed weights model from a geopolitical rival.
That's a very reasonable stance. It doesn't change the fact that we do have plenty of local models (up to and including Qwen 3.5) that are still quite useful.
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 open weight, but I'd love to sub a decent competitive coding plan from a European or Chinese provider. Right now they're not quite there. If closing it and charging for it brings them closer to competitive, that's ok.
If the US tech and AI industry long term wants customers and a broad market outside of their own domestic base, they need to reconsider who they are bending the knee to, and how they are defining their policies in relation to the Trump administration.
Bring on the Chinese competition.
Ah, so that explains the recent wave of Qwen team-member departures.
I’m starting to wonder where the most is for any of these models.
Sure they are not cheap to train. But if open weight models continue to be trained and continue to become available on cheaper hardware, how do dedicated AI companies protect their margins?
> Initial reactions are mostly anger from everyone who didn’t realize that the play along was to give away the smaller models as advertising, not because they were feeling generous.
The naivety around this has been staggering quite frankly. All of a sudden, people thinking that meta etc are releasing free models because they believe in open access and distribution of knowledge. No, they just suck comparatively. There is nothing to sell. Using it to recruit and generate attention is the best play for them.
I don't think there's so much naivety. People can be aware of the the plan and still be frustrated and disappointed when it happens.
For a brief moment there were a lot of comments about how Chinese tech companies are our saviors in the age of AI because they were releasing their models. It was an edgy contrarian take that was getting a lot of traction, mostly from commenters who were unfamiliar with Alibaba and thought it was the anti-Big-tech
I'm not frustrated or disappointed, we have lots of models from Qwen already. We haven't really lost anything. And plenty of players only release "smaller" models anyway, so it's hardly unprecedented.
I thought Qwen was releasing open-weight because China can't compete with America (because of people's privacy concerns), so the only thing they could do is salt the ground economically with open models, and make sure everybody loses.
Apparently that wasn't actually the play here.
Qwen is actually a pretty strong player in the Chinese market. There is an implied "salt the ground" play but it's mostly from hardware makers, who are trying to keep the big AI players honest and also stand to gain if local inference becomes popular.
Opus was released in Feb 2026. Even though it feels like a long 2 months has passed, its' not really clear that they were developing this as a competitor to that product.
There's nothing really strange about not competing directly with the best, but rather showing whom you are as good as.
I don’t know why anyone would do the mental backflips to defend this.
They posted charts with logos for Claude and others. You had to read the fine details to realize they weren’t comparing to the latest offerings from those companies. They were counting on you not noticing.
There’s zero reason to compare to old models unless you’re trying to mislead.
> I think there is a moderately large market for models like this that aren’t quite SOTA level but can be served up much cheaper.
There isn't, pretty much everyone wants the best of the best.
The OpenRouter usage stats indicate the opposite: https://openrouter.ai/rankings?view=month
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 those APIs directly.
I use ChatGPT and Claude on OpenRouter, because it's just easier than buying credits on each platform separately.
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.
what happened around jan this year(26) that caused such a climb in usage?
Openclaw
No. Right now I'm upset that Google has removed (or at least is in the process of removing) the Gemini 2.0 flash model. We use it for some pretty basic functionality because it's cheap and fast and honestly good enough for what we use it for in that part of our app. We're being forced to "upgrade" to models that are at least 2.5 times as expensive, are slower and, while I'm sure they're better for complex tasks, don't do measurably better than 2.0 flash for what we need. Yay. We've stuck with the GCP/Gemini ecosystem up until now, but this is kind of forcing us to consider other LLM providers.
this is one of the reasons im hearing more and more people are using open/locally hosted models. particularly so we dont have to waste time to entirely redo everything when inevitably a company decides to pull the rug out from under us and change or remove something integral to our flow, which over the years we've seen countless times, and seems to be getting more and more common.
products entirely disappearing or significantly changing will be more and more common in the llm arena as things move forward towards companies shutting down, bubbles deflating, brand priorities drastically reshifting, etc...
i think, we're at or at least close to a time to really put some thought into which pieces of your flow could be done entirely with an open/local model and be honest with ourselves on which pieces of our flow truly needs sota or closed models that may entirely disappear or change. in the long run, putting a little bit of thought into this now will save a lot of headache later.
What’s interesting about this is that for previous technologies you could define a standard and demonstrate compliance with interfaces and behavior.
But with LLMs, how do you know switching from one to another won’t change some behavior your system was implicitly relying on?
In case you don't know, Gemini 2.5 flash is hosted on DeepInfra. They also have 1.5 flash but not 2.0 flash.
I have no affiliation with DeepInfra. I use them, because they host open-source models that are good.
> There isn't, pretty much everyone wants the best of the best.
For direct user interaction or coding problems, perhaps. But as API calls get cheaper, it becomes more realistic to use them for completely automated workflows against data-sets, or as sub-agents called from expensive SOTA models.
For example, in Claude, using Opus as an orchestrator to call Sonnet sub-agents, is a popular usage "hack." That only gets more powerful, as the Sonnet equivalent model gets cheaper. Now you can spawn entire teams of small specialized sub-agents with small context windows but limited scope.
That is a very complex, high level use case that takes time to configure and orchestrate.
There are many simpler tasks that would work fine with a simpler, local model.
Exactly.
I did create my own MCP with custom agents that combine several tools into a single one. For example, all WebSearch, WebFetch, Context7 exposed as a single "web research" tool, backed by the cheapest model that passes evaluation. The same for a codebase research
Use it with both Claude and Opencode saves a lot of time and tokens.
> But as API calls get cheaper, it becomes more realistic to use them for completely automated workflows against data-sets
Seems like a huge waste of money and electricity for processes that can be implemented as a traditional deterministic program. One would hope that tools would identify recurrent jobs that can be turned into simple scripts.
It depends on the specific task.
For example: "Here our dataset that contains customer feedback comment fields; look through them, draw out themes, associations, and look for trends." Solving that with a deterministic program isn't a trivial problem, and it is likely cheaper solved via LLM.
Ever hit your daily limit on Claude Code and saw how expensive it is to pay per token?
maybe there isnt, but as understanding grows people will understand that having an orchestration agent delegate simple work to lesser agents is significant not only for cost savings, but also for preserving context window space.
For coding I want the best. Both I and $work do lots of things besides coding where smaller models like qwen3.5-27b work great, at much lower cost.
That isn't true. In a Codex or Claude Code instance, sure... but those are not the main users of APIs. If you are using LLMs in a service for customers, costs matter.
The market for API tokens is bigger than people like you and I (who also want the best) using then for code.
There are a lot of data science problems that benefit from running the dataset through an LLM, which becomes bottlenecked on per-token costs. For these you take a sample subset and run it against multiple providers and then do a cost versus accuracy tradeoff.
The market for API tokens is not just people using OpenCode and similar tools.
Nope. I get very good results from GLM 5 and 5.1. I’m not working on anything so complex and groundbreaking that I need the best.
Coding is a rung on the ladder of model capability. Frontier models will grow to take on more capabilities, while smaller more focused models start becoming the economical choice for coding
Everyone may want the best, but the amount of AI-addressable work outstrips the budget available for buying the best by quite a wide margin.
OpenCode allows for free inference tho.
Not really. It depends on the usecase. For private stuff I'm very happy to take what was SOTA a year or 2 ago if I can have it all running in my home and don't have to share any of my data with some sleazy big tech cloud.
The price is a concern too of course. But privacy is a bigger one for me. I absolutely don't trust any of their promises not to use data for training purposes.
That's only because current models don't saturate people's needs. Once they are fast and smart enough people will pick cheaper ones.
How stupid somebody has to be to mix up Opus with Qwen?
OP didn't say about confusing Opus with Qwen but rather people being confused about Qwen3.6-Plus not being available as an "open weight" model available for self hosting.
I understand peoples reactions of Qwen team comparing against Opus 4.5 instead of 4.6. And them comparing against Gemini Pro 3.0 instead of 3.1. But calling it misleading is a bit of stretch in my eyes, people here are acting like we immediately forgot how previous generations performed just because a new version is released.
This field is going in a incredible pace, the providers release a new model every quarter or so. The amount of criticism is a bit overblown in my opinion. The benchmarks still look very good to me. I’ve used GLM-5 (latest is GLM-5.1) and Kimi K2.5, they are decent and gets the job done, so seeing how this model of Qwen performs compared to it is kinda impressive.
Also, why are so many pointing out the fact that this model is not open-weight as if this is their first time doing so. Qwen-3.5-plus, Qwen-3-Max is also closed source. This is not something new.
I think Qwen trying to catch up to the SOTA models is still healthy for us, the consumers. Sure, its sad news that this version is closed-weight, but I won’t downplay their progress.
Opus 4.5 is already pretty good.
Opus 4.5 is $25/m output tokens.
This is at most $6/m output tokens.
That's ~1/4 the price.
I think it’s more the principle of deception that upsets people. Imagine if Apple released a new iPhone and publicly compared its specs to some previous gen Android. It’s not in good faith.
They compared their M-series chips to older Intel Macs for a while, likely to target users who were still on Intel chips. If they released a lower cost iPhone and compared it to a previous gen Android I could see the reasoning for it. It's not deception if it's a valid comparison and people just fail to understand what's being compared.
Now, is it mildly deceptive because all of the companies using incredibly confusing naming conventions for their models? Maybe!
Apple continues to compare to prior versions of Apple Silicon. I suspect it is a mix of trying to provide useful, realistic upgrade information and numbers that still sound good for those not paying attention.
I don't think any org doing this is necessarily being deceptive, so long as there's some reasonable basis for the chosen comparable(s).
For example, comparing a new iPhone to a prior Android phone might make sense if the install base is considerably large and Apple is targeting the cohort for user acquisition. (~"These benchmarks are not for you.")
The community will always run the numbers and get the clicks for the benchmarks not filled in by the 1st party. I noticed what appeared to be some movement from Apple in content they've produced to get ahead of this with recent product content.
Why are we so quick to call it deception? Their figure is quite clear. They aren't fiddling with the graph or hiding the labels, they are clearly stating which models it compares against. But I agree on the sentiment that the standard practice should be to bench against the latest SOTA models.
Even if openly stated, why would they be comparing to a previous generation if not for deception?
Laziness? Lack of time? It's not like the latest generation of the SOTA models were released yesterday.
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!
Pelican is drafting rear peloton
Worth noting that this model, unlike almost all qwen models, is not open-weight, nor is the parameter count exposed. Also odd that it is compared against opus 4.5 even though 4.6 was released like 2 months ago.
They said in the last paragraph[0]:
"[...] In the coming days, we will also open-source smaller-scale variants, reaffirming our commitment to accessibility and community-driven innovation. [...]"
[0] https://qwen.ai/blog?id=qwen3.6#summary--future-work
> we will also open-source smaller-scale variants
In other words, like GP said, this Qwen3.6-Plus model is not open-weight unlike the other Qwen models.
Qwen3.5-Plus is the largest variant of the open weight Qwen3.5 model, expanded with a 1M context window and fine-tuned on the Qwen-native harness’ specific tools.
In a practical sense, I'm primarily interested in small to medium sized models being open. I think that might be common sentiment.
However, my hope is that there will be at least somewhat competitive big and open models as well, from an ethical/ideological perspective. These things were trained on data that was provided by people without their consent, so they should at least be be publicly accessible or even public domain.
> unlike almost all qwen models
Almost all means there have been ones before that were not open. So, no contradiction there.
> unlike the other Qwen models
Please send the download link for qwen 3.5-plus.
Also, who cares? If you have the hardware to run a ~400b model i don’t think you count as a home user anymore.
If Opus 4.6 was only released two months ago, then it seems reasonable that Qwen hasn't finished fully comparing against the latest Opus.
I wouldn't say "almost all" seeing as -MAX and -Omni models were always closed.
I'll diverge from some of these comments, I don't find it misleading to compare to Opus 4.5.
I can remember how good Opus 4.5 was. If I'm considering using this, it's most informative to me to compare to the model it's closest to that I have familiarity with.
I'm obviously not switching to this if I want the best model. I'm switching if I'm hopeful that the smaller versions are close to it, or if I want to have more options for providers, or for any other reasons unrelated to getting the highest quality responses possible.
Exactly this. If you can get something close to Opus 4.5 for free, that's noteworthy. I may not use it for the most critical pieces of my app, but not everything I do is galaxy-brain coding.
Yes, honestly, Opus 4.6 and GPT 5.4 were mostly not really noticeable improvements over 4.5 and 5.3 respectively. If we were stuck at 4.5 levels but at 1/10th of the price, I'll take it.
I find 4.6 pretty noticeable upgrade, but it might be the 1M context. I'm interested in how the 1M context works out with Qwen.
I’m surprised that people are surprised. Qwen has been hosting private plus and max variants for a while now.
Just more evidence that the B tier models are six months behind. Ultimately that’s good. Opus 4.6 level intelligence will be cheap later this year!
The benchmarks provided are for Opus-4.5, not for the latest Opus-4.6 and Qwen is still lagging in a lot of them.
There is no reason to benchmark against Opus 4.5 when Opus 4.6 has been out so long, other than to be misleading.
I can see reasons, among others that 4.5 was the one established as they were preparing this version. "So long" is merely 2 months ago, and Qwen 3.5 was barely released less than 2 months ago. They were likely already working on finalizing 3.6 before 3.5 official launch, and as 4.6 came out.
In any case, aside Claude fanboyism, having other plays inch closer to similar performance is always useful. Even if they are "6 months behind" as the pace slows down, this guarantees that there's no huge moat and they'll eventually either get to where the SOTA is, or the difference wont be that big.
I'd rather put fewer eggs in 2-3 big player baskets.
And it seems they've decided to go closed-source for their largest, best models.
3.5-plus was also only available via api. I don’t know what the long term business model for open weights is, I hope there is one, but it seems foolish to assume that companies will be willing to spend millions of dollars of compute on an asset worth zero in perpetuity.
The business case is to salt the earth for new competitors, coupled with marketing.
They always did that. Did they say anywhere they'd open all their models? They still have a business.
They've always had closed-source variants:
- Qwen3.5-Plus
- Qwen3-Max
- Qwen2.5-Max
etc. Nothing really changed so far.
> In the coming days, we will also open-source smaller-scale variants, reaffirming our commitment to accessibility and community-driven innovation.
Looking forward to when this gets on Bedrock. I built an app with a niche AI agent and to this point only Sonnet is really good enough for our use case, but its expensive!
Try using Grok 4.1 reasoning. It's crazy cheap, and really it's not that bad.
I hope their open source variants are just as good, having a 1 million token window for a fully offline model would be VERY interesting.
I don't know how well it performs, but you can extend Qwen3.5 to 1 million token context using YaRN. Also, Nemotron 3 Super was recently released and scales up to 1 million token context natively.
The agent benchmarks here are interesting but I'd love to see how Qwen3.6-Plus handles long-horizon tasks where it needs to recover from its own mistakes. Most agent evals test the happy path. The hard part is when the model takes a wrong action at step 3 and needs to recognize and backtrack at step 15. Has anyone stress-tested this in a real dev workflow?
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.
I would love to hear from people using both (Claude Code OR Codex) AND (Qwen) and their experience with Qwen models, are they on par, or how far are they?
I switch between Claude Code (Opus/Sonnet) and Qwen (OpenCode, OpenClaw) multiple times throughout the day and Qwen 3.5 is really nice. I do also use KimiK2.5 and GLM5 pretty often too and I'm starting to get a sense that the agent tool is becoming a little more important than the model with these level of models. As long as tool calling and prompt quality is all configured correctly by the provider.
It is no longer available on OpenRouter. They say "going away on 3-March", but it's already gone!
Nice, I hope there will also come a small open version of it.
How convenient of them to compare themselves to the last generation Opus and GPT models to make their model look better than it really is.
It's not open weights so I'm not interested.
Does anyone have experience with Alibaba's coding plan? Not that I'm very tempted at $50/month...
A bit off-topic but I’m on the legacy Lite plan (now discontinued), and it’s more than enough for hobby projects. The main draw is the generous request-based quota (18k requests/month) rather than a token-based one.
This means a 100k token request counts the same as a 100-token one. I’ve made about 8000 requests in the last two weeks, averaging around 80k tokens per request. It feels like they’re subsidizing this just to gather data on agentic workflows.
On the downside, the speed is mediocre (15–30 tg/s for GLM-5), and I’ve seen the model glitch or produce broken output about 10 times out of those 8k requests.
Quite strong results in the benchmarks but why Gemini 3 Pro instead of 3.1? Why only for a few of the benchmarks? Why is OpenAI not there in the coding benchmarks? Why Opus 4.5 and not 4.6? Just jumps out into my eye as a bit strange.
As always, we'll have to try and see how it performs in the real world but the open weight models of Qwen were pretty decent for some tasks so still excited to see what this brings.
the comparison is helpful but i'd want to see how it handles edge cases
Not really interested in using models hosted on alibaba cloud.
Like Qwen local for it’s privacy, but I trust the privacy of Google/OpenAI/Anthropic more than alibaba.
I had the exact opposite reaction. I stopped using OpenAI/Google a while ago due to privacy and moved to local Qwen, now I'm considering using Alibaba cloud. You know Google and OpenAI are going to share everything with the US government and Western ad networks. But with Alibaba, who cares if the CCP & Chinese ad networks have a comprehensive profile on me? From a pragmatic perspective it's much better for (outcomes related to) privacy.
so if China has the data good, us has the data bad, got it lol.
us actually has laws around this and they arent sharing very much with thr us gov today. china shares 100% as required by law. and neither care much about "how long do i cook eggs for", but they do care about code generation a lot.
From an espionage perspective your own government is the safest. But from a civil rights perspective your own government is your most immediate threat. China isn't going to arrest me for my opinions on Netanyahu, my own government could
And the US government has repeatedly shown that it is very interested in collecting all the data available, just like China. In China this is simply done in the open while the US has a veneer of protection for citizens. But where the data collection is forbidden by law they either ignore the law or ask another five eyes member to do the spying and share the results. Both are well documented
> China isn't going to arrest me for my opinions on Netanyahu, my own government could
I don't know whether you really believe this or it was an off the cuff remark. China is not going to tell you why they plan to arrest you. China is not a benevolent dictatorship.
> so if China has the data good, us has the data bad
It's not that, it's about relative risk to your own life. Asking questions about "DEI" for example is much more likely to have adverse effects on your life if you ask Grok or an OpenAI chatbot, though still not that likely.
So I guess if it’s your personal data that’s up to you, but if you have private client data and client of your client data and that’s the fundamental reason why you are doing local ai, I can’t imagine moving from qwen local to qwen alibaba after not choosing google/anthropic/openai
As with all arguments equivalent to "I have nothing to hide, so I have nothing to fear," it may be true now, but it may not be true later. The only certainty is that this will not be your call.
Agreed
> Like Qwen local for it’s privacy, but I trust the privacy of Google/OpenAI/Anthropic more than alibaba.
None should be trusted, unless you are running them locally.