Rendered at 16:12:47 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
gvkhna 11 hours ago [-]
If it’s not zhipu then why is it returning errors that zhipu does for other models? Who else would return the exact same errors even if they took a lot of core infra like tokenizer from z?
e9 10 hours ago [-]
Someone could've trained model on top of GLM. Same way Cognition trained their SWE model on top of Kimi and Cursor did same with their Composer model.
gvkhna 10 hours ago [-]
While possible the amount of variation in serving infrastructure is unlikely to land with actually giving the exact same errors zhipu does.
It feels like glm flash, and there was a report zhipu had secured a huge new cluster suggesting they have the capacity. My guess anyway.
The reasoning levels are the same as GLM 5.3. GLM 5.3 is still not open...
I believe it's GLM 5.3 Flash or Air.
weiran 10 hours ago [-]
Reasoning levels are often just injected system prompts so not a great way to fingerprint models.
ggcr 10 hours ago [-]
Ziphu has that many resources to be able to serve capacity for 1 quadrillion tokens per day on Nous portal? My bet is that it's a Composer model from Cursor running on xAI cluster, they already did a Composer based on Kimi-K2.5
LaurensBER 9 hours ago [-]
There's three options here:
- The provider has a massive amount of (unused) hardware. Google or Cursor seem most likely
- The model is extremely efficient, beyond anything we've seen so far
- Whomever made the model has improved the cache efficiency in such a way that it's very cheap to serve. See e.g Deepseeks or Xiaomi caching (pre-price increase)
re-thc 9 hours ago [-]
Option 4: the claimed capacity is not true. Real world usage hasn’t reached anywhere close to it.
johndough 8 hours ago [-]
> 1 quadrillion tokens per day on Nous portal
If you are referring to this number (https://xcancel.com/NousResearch/status/2090899914700054780), they are either mistaken, or they mean that they can route 1 quadrillion tokens per day, but the provider behind Ox Alpha certainly can't provide that. Almost all of my requests have hit a rate limit so far.
SyneRyder 7 hours ago [-]
I was hitting 429 overloaded regularly with Ox on OpenRouter yesterday, but a lot of that turned out to be problems with my harness. I fixed some bugs, improved the back-off, and I haven't hit a 429 error since (touch wood).
OpenRouter says they're doing 6 Trillion tokens a day with Ox Alpha so far, and it has been their biggest launch of all time. OpenCode claimed they had capacity for 100T a day.
GLM 5.3 and all previous models don't have a vision encoder and can only accept text. Ox-Alpha can accept video and images, so unless Z-ai added a pretty good vision encoder for this model, I don't think so.
My money is on Moonshot and this being Kimi K3.5. The measured tps and latency is in-line with K3's tps and latency from Moonshot.
MiniMax M3.5 is also possible (but the MiniiMax provider is a lot more performant than the lab behind ox-alpha, so less likely).
nylonstrung 21 hours ago [-]
It would be stranger to me that Kimi switched to GLM's tokenizer than that GLM added multimodal like Kimi and Deepseek both did recently
Bolwin 22 hours ago [-]
Glm had made vision models in the past. Look up GLM 5v.
The only question now is if it's 5.3v, 5.4/5.5 or a dedicated flash/vision model
BoredomIsFun 11 hours ago [-]
GLM made pretty decent for that time small 9b vision model, GLM-4.1.
volf_ 22 hours ago [-]
Yeah. It could be. The Z.ai DC latency is still ~1.2s faster than whomever is serving this model.
Almondsetat 22 hours ago [-]
DeepSeek literally just came out with the vision-enabled version of Flash v4 which was purely text based. Why would GLM not be able to do the same thing?
volf_ 22 hours ago [-]
It's possible
minimaxir 12 hours ago [-]
The other tell from the provider angle is capacity. Whoever is hosting Ox Alpha has a lot of capacity which narrows down a lot of the Chinese companies.
Havoc 9 hours ago [-]
> and all previous models
They have had vision models before just not their flagships
ggcr 9 hours ago [-]
Z.ai founders were one of the pioneers in MM-LLMs with Cog-VLM years ago, back when LLaVA emerged. I wouldn't be surprised if they added multi-modal capabilities
TiredOfLife 10 hours ago [-]
But Moonshot limited signups because they lacked compute.
The upper chart shows the availability of Ox Alpha and the lower chart shows the availability of GLM 5.3 by Z.ai. They had a blip at exactly the same time.
troysk 8 hours ago [-]
It is probably from Google and is probably hosted on Vertex AI. Opencode announced that responses from Ox Alpha should be better and soon posted about Vertex eu and us multi region update in their changelog. I could be a Gemini model or a new one based on GLM based on tokenizer. Also the amount of inference it is providing for free is something only google can support with its TPUs. So maybe a GLM based model running on TPUs.
walrus01 7 hours ago [-]
I've seen a number of people report that it answers near-identically to mainland CN built models on topics related to controversial things the CCP doesn't want to talk about. I'd be extremely surprised if it's a Google model.
dezgeg 3 hours ago [-]
I extremely doubt it's Gemini - all previous Geminis have had very different output style from other models. In a coding harness Gemini is extremely quiet and mostly just does thinking and tool calls, with the thinking blocks heavily summarized.
Also Claude and the Chinese models really like to say "Let me X", Geminis preference is "I will X".
jerrythegerbil 11 hours ago [-]
As someone who uses NCD nearly every day, I have concerns about how it’s been used here.
But while we’re “guessing”: Xiaomi MiMO
dejanseo 6 hours ago [-]
Author here. Tell me more. How do you use NCD and what is your concern about its application here? Should I have use da larger reference sample? You can test it in action here: https://dejan.ai/tools/ai/ (e.g. drop a claude article or GLM article in and see what it says, it's not perfect but reasonably good).
walrus01 11 hours ago [-]
Have also seen people guess it's a next version of Longcat, but I also think that's unlikely
pijalu 9 hours ago [-]
My bet: it's Google running a "new" model based on GLM
Zylokloto 4 hours ago [-]
That would be a complete defeat for a Frontierlab / Company like Google/Deepmind.
They will never ever do like this.
They for sure would add architecture ideas from other research/models but thats it.
simianwords 9 hours ago [-]
Interesting. I like this theory because it explains why some G employees were vague posting about it. But like.. why? Why wouldn’t Google just use Gemma?
re-thc 8 hours ago [-]
> why some G employees were vague posting about it
They were just trolling.
Otterly99 9 hours ago [-]
The Ox-Alpha webpage really make it sound like they are trying to hype a model that has nothing particular to show:
"The reasoning model that appeared out of nowhere. Built for code, long-horizon agents, and a million tokens of context. Nobody knows who made it — everyone wants to try it."
10 hours ago [-]
tadkar 11 hours ago [-]
I wonder if the NCD metric says something about distillation too. Would you expect that a model that has been distilled/seen traces from other models would have a smaller NCD? It would be really interesting to see if this holds up and provides evidence of distillation or certainly evidence of model outputs being used in the training mix.
dejanseo 6 hours ago [-]
Hello! I wrote the above article and did the NCD on model outputs. The same thought crossed my mind when I saw Gemma misclassified as Gemini quite frequently. And GLM almost as Claude and not as Gemini at all. Gave me the feeling as if GLM didn't train on Gemini generated synthetic data at all but mainly on Claude and GPT.
Its amazing to me that providers haven't added any sort of masking of the prompt in the thinking traces to avoid prompt extraction via this sort of trivial attack
dejanseo 6 hours ago [-]
Hey, it may be trivial, I'm pretty proud of it :)
simianwords 9 hours ago [-]
My strong prediction: this model is around 64B and can run on laptops. Thats the reason behind the hype.
anon373839 9 hours ago [-]
It's fun to imagine that it could be GLM 5.3-Flash. Between GLM 4 and 5, the flagship's total parameters doubled and the active parameters went up 25%. GLM 4.7-Flash was 30B / 3B active. If this model were 60B / 4B active, that sure would hit a sweet, currently empty spot in the lineup of open models.
But I don't know. There are vagueposts on X about this model running on two DGX Sparks. If they did the same scale-up with Air instead of Flash, that would probably be about right.
re-thc 8 hours ago [-]
> this model is around 64B and can run on laptops.
We're lucky if it'd fit in 1 DGX Spark. Laptops - nah, unless you mean like an M5 Max with 128GB of RAM then maybe.
> Thats the reason behind the hype.
The hype is imagine DeepSeek Flash before the price increase with even better performance. It'd be like unlimited Sonnet.
mogili 11 hours ago [-]
It's not a good model tbh, got a bunch of things wrong that Opus corrected in my codebase.
gvkhna 10 hours ago [-]
The harness is making a big difference, lackluster performance with pi but somehow very good performance with opencode. There’s some rl there for sure, for a smaller model it’s likely going to perform much better in a harness it understands the best.
SyneRyder 7 hours ago [-]
I've been using a harness that I made myself, and making improvements to the harness has vastly improved its performance. It's actually been a useful model to identify flaws in the harness.
I think something also went wrong with the Ox provider last night (at least on OpenRouter), for a few hours it wouldn't accept tools. Zero change to the harness while I slept and it was back working again the next morning.
petesergeant 11 hours ago [-]
Yet to find a model that cross-model review doesn’t find a bunch of things wrong with. I’m running simultaneous review with whichever of Grok4.6/GLM5.3/Fable/Sol didn’t write it, and each model tends to find items the others didn’t.
mogili 10 hours ago [-]
Wasn't just a review, it failed the task I gave and Opus completed the task
epolanski 10 hours ago [-]
If your changes are non trivial even the same model will loop over and over with the feedback.
Nvidia is quite open in publishing and talking about their models though.
petesergeant 11 hours ago [-]
I think within 12 months we’re going to see a frontier (inc open models) that’s so good at almost all human-directed tasks that which model you use just won’t matter. Only differences that remain will be in deep research or very long-range tasks.
stingraycharles 11 hours ago [-]
People were saying this last year, and they’ll be saying the exact same thing next year. The goalpost keeps moving.
Tepix 10 hours ago [-]
It‘s already happening, people are using cheaper models because they are good enough
petesergeant 10 hours ago [-]
Someone else having been too early on a prediction has little bearing on my prediction. A year ago almost nobody was using open models as daily drivers, today they are. When I run out of Fable and Sol credits in a week, I switch to GLM5.3, and it's not quite there, but it's good enough for productive work.
vladimirmackic 5 hours ago [-]
[dead]
behnamoh 11 hours ago [-]
[flagged]
walrus01 11 hours ago [-]
I wasn't aware that an inanimate piece of software run by a corporation can be 'doxxed'. Totally inaccurate use of the neologism.
minimaxir 11 hours ago [-]
You cannot "dox" an AI model.
Given the traction the model has received, it is extremely newsworthy to know who's developing and hosting it.
behnamoh 11 hours ago [-]
my question is: how does that affect a company's strategy? it's not like management is gonna switch models soon as a new shiny one drops. entire workflows depend on specific models working the way they do; you can't just swap out models.
minimaxir 11 hours ago [-]
If it's a really really good model, then yes, people will switch as long as the price is right. Ox Alpha is looking to be a really really good model to the point that it competes with Fable/Sol, and will likely beat them on price.
volf_ 11 hours ago [-]
same business model as crack. best way to get people hooked is to make the first hits free.
JSR_FDED 11 hours ago [-]
You can learn from the variety of techniques they used to come to this conclusion.
It feels like glm flash, and there was a report zhipu had secured a huge new cluster suggesting they have the capacity. My guess anyway.
https://www.tomshardware.com/tech-industry/artificial-intell...
I believe it's GLM 5.3 Flash or Air.
- The provider has a massive amount of (unused) hardware. Google or Cursor seem most likely
- The model is extremely efficient, beyond anything we've seen so far
- Whomever made the model has improved the cache efficiency in such a way that it's very cheap to serve. See e.g Deepseeks or Xiaomi caching (pre-price increase)
OpenRouter says they're doing 6 Trillion tokens a day with Ox Alpha so far, and it has been their biggest launch of all time. OpenCode claimed they had capacity for 100T a day.
https://x.com/OpenRouter/status/2091912024922177562 https://x.com/opencode/status/2090544355824038300
My money is on Moonshot and this being Kimi K3.5. The measured tps and latency is in-line with K3's tps and latency from Moonshot.
MiniMax M3.5 is also possible (but the MiniiMax provider is a lot more performant than the lab behind ox-alpha, so less likely).
The only question now is if it's 5.3v, 5.4/5.5 or a dedicated flash/vision model
They have had vision models before just not their flagships
https://files.catbox.moe/k52n6k.png
The upper chart shows the availability of Ox Alpha and the lower chart shows the availability of GLM 5.3 by Z.ai. They had a blip at exactly the same time.
Also Claude and the Chinese models really like to say "Let me X", Geminis preference is "I will X".
But while we’re “guessing”: Xiaomi MiMO
They will never ever do like this.
They for sure would add architecture ideas from other research/models but thats it.
They were just trolling.
"The reasoning model that appeared out of nowhere. Built for code, long-horizon agents, and a million tokens of context. Nobody knows who made it — everyone wants to try it."
Its amazing to me that providers haven't added any sort of masking of the prompt in the thinking traces to avoid prompt extraction via this sort of trivial attack
But I don't know. There are vagueposts on X about this model running on two DGX Sparks. If they did the same scale-up with Air instead of Flash, that would probably be about right.
We're lucky if it'd fit in 1 DGX Spark. Laptops - nah, unless you mean like an M5 Max with 128GB of RAM then maybe.
> Thats the reason behind the hype.
The hype is imagine DeepSeek Flash before the price increase with even better performance. It'd be like unlimited Sonnet.
I think something also went wrong with the Ox provider last night (at least on OpenRouter), for a few hours it wouldn't accept tools. Zero change to the harness while I slept and it was back working again the next morning.
Ox Alpha
https://news.ycombinator.com/item?id=49381896
Given the traction the model has received, it is extremely newsworthy to know who's developing and hosting it.
Yet here you are, reading AND commenting about it