Despite what any benchmarks tell you, I'm actually finding GLM-5.3 max to be better than Sol and Fable.

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 better than anything we're getting access to.

Weights on HF here: https://huggingface.co/zai-org/GLM-5.3-Flash

I decided to take the plunge and get myself four sparks at a decent price (and bought the QSFP cables from AliExpress because they are literally 1/2 the price of Amazon), even knowing Apple was going to release new hardware and there's probably a spark 2 on the horizon. It looks like this is going to be a decent fit for what I need. I've been experimenting with a two-node DS4 and it's _good_ at some tasks, but it really just spins its wheels when it hits the limit of what it can reason through.

I can offload mundane/basic tasks to DS4 on two sparks, but I've been pushing it harder on some novel work and it just can't run on its own at all beyond a certain complexity level.

I would love to see an Opus-4.8-level local model but TBH I just haven't got there yet. The models I've tried so far _are_ good but they aren't able to solve tough technical challenges, regardless of harness/prompting/etc.

This is going so fast! What a time to be on hackernews:

July 16th: The "Kimi K3 moment" - China has caught up to Opus!

4 weeks later: GLM 5.3 - Same performance, but cut the amount of parameters and cost to a third!

12 days later: GLM 5.3 Flash - Almost GLM5.3 performance but cut the parameters in half, cut prices to a fifth and serving on Chinese chips!

Chinese labs are so used to manipulating benchmarks to try to flatter inferior models that when they finally have one that's really pretty good I think the official announcement here undersells it.

https://deepswe.datacurve.ai/

That's pretty solid. Smarter and cheaper than Luna xhigh, not as smart but less expensive than Luna max. Smashes deepseek v4 flash, and even worse it matches v4 pro at a tiny fraction the cost. Roughly equivalent to sol medium, at a fraction the cost.

They should've just lead with real, up to date data, because it's good, not the silly old tactics like comparing to Opus 4.8 when 5.0 is out in many of their charts.

Congrats to them!

You guys read Z.ai's terms of service, right?

Broad and perpetual license over inputs and outputs, and even your name and profile picture.

Vague prohibitions on whatever may harm Z.ai’s "interests" or even the "national interests" of any country.

Vague prohibitions on "disturbing" or "inappropriate" content, whatever that is.

Vague prohibitions on discussing Z.ai, even my posting this comment violates it.

Can ban you if you, in the "sole and absolute opinion" of Z.ai, have violated these broad terms, and if you paid for the discounted yearly plan kiss your money goodbye.

If you're on opencode's go $10/mo plan and want to use GLM-5.3-flash right now on pi, you can add this to models.json until pi updates to support it:

    {
      "providers": {
        "opencode-go": {
          "models": [
            {
              "id": "glm-5.3-flash",
              "name": "GLM-5.3 Flash",
              "api": "openai-completions",
              "baseUrl": "https://opencode.ai/zen/go/v1",
              "reasoning": true,
              "input": ["text", "image"],
              "cost": {
                "input": 0.15,
                "output": 0.5,
                "cacheRead": 0.03,
                "cacheWrite": 0
              },
              "compat": {
                "supportsStore": false,
                "supportsDeveloperRole": false,
                "maxTokensField": "max_tokens"
              },
              "contextWindow": 1000000,
              "maxTokens": 131072,
              "thinkingLevelMap": {
                "off": null,
                "minimal": null,
                "low": "low",
                "medium": null,
                "high": "high",
                "xhigh": null,
                "max": "max"
              }
            }
          ]
        }
      }
    }

> Before release, we tested GLM-5.3-Flash anonymously as ox-alpha on OpenCode and OpenRouter to gather user feedback. It quickly became the most popular model of the week — with all of this traffic served on Chinese AI chips.

Just like that we are witnessing an open burial. It's now in everyone's interest to keep the valuations in the 'A.I' economy as they're though it's apparent they're not justified.

whether it's the cost to develop models, cost of hardware, cost of serving ie inference.

> with all of this traffic served on Chinese AI chips

RIP Nivida shareholders

Google was so ahead when it made this statement: "We Have No Moat And neither does OpenAI" - May 04 2023

https://newsletter.semianalysis.com/p/google-we-have-no-moat...

Correction: Not a statement, rather, an internal memo by a Google employee. Thanks @granzymes for highlighting.

although i initially thought it didn't make sense financially to run this kind of model locally, i did run the numbers and for heavy users this could justify buying $10k worth of hardware with a ROI over a few months, less than a year.

I was looking at my token usage, mostly from subsidized codex/grok subscriptions and i'm a somewhat heavy user. The thing is i would actually use even more tokens if it wasn't for the weekly quotas.

In the end, with a $10k investment and running this kind of model, estimating a 2x increase in token usage because i wouldn't have weekly quotas and comparing to glm api prices, this thing could pay for itself in less than a year.

Obviously i'm paying subscription price right now, so the math doesn't work. Although using local ai removes all weekly quotas. Keep a subscription to have access to frontier models for planning work, and local hardware + glm-5.3 flash for implementation, e2e testing, qa work 24/7.

It's not that crazy of an idea and the numbers aren't that bad.

So the vagueposting by googlers about Ox Alpha was just... what exactly?

Like I get that they have to be careful about comms, but surely senior members of the team can clarify when something is NOT them, when everyone is gosspiing it is them.

> 320B total parameters and just 18B active parameters

This is pretty hefty for a "flash" model, even a 256 GB setup is insufficient at q4 - and q4 is already the worst-but-still-acceptable quant in my experience. The benchmarks look great, especially since GLM tends to be more honest than the average Chinese lab, but you’ll need to splurge to run it at home.

@edit: so many releases that I forgot to math. This fits just fine in q4, realistically the minimal hardware would be 192gb - so blazing fast on double rtx 6000 pro and usable on 256gb unified memory. You could even go with 5bit quant on 256gb.

… you’ll still need to splurge, though.

If we fast forward say 5 years, I don't see how we don't end up in world where people (and enterprises) are more savvy with how they use LLMs. Meaning, more models, smaller models, weirder models, more specialized models, etc. And all of it running on a variety of hardware (edge devices, personal computers, on-demand cloud compute).

I don't see how NVIDIA can keep their spot as belle of the ball. If LLMs and friends are truly to become as useful and ubiquitous as everyone thinks they will, then commoditization is the only option.

Nice, finally they fixed the huge reasoning tokens count.

Now it's similar cost to DeepSeek v4 flash, but smarter.

My tests: https://aibenchy.com/compare/z-ai-glm-5-3-flash-max/deepseek...

With tiny models surpassing huge, 6 months old models on benchmarks, does anybody have some smart words to share on how these still "feel" different?

Artificial Analysis ranks GPT 5.6 Luna similar to GPT 5.4, but that never matches my real world experience. AA seems to do a good job making a single number as representative as possible but there is still so much benchmarks don't communicate.

> it outperforms GLM-5.2 across benchmarks and real-world workloads at one-tenth the price, while approaching Claude Opus 4.8 on coding and agentic benchmarks.

From a biased source, but would be big if true. I've had great results with GLM 5.2.

From their subscription page, the smallest plan gives you about 97M tokens weekly for 5.3 but 292M for 5.3 Flash. Not exactly 10x the limit.

For those who didn't read, this is the identity of the mysterious "Ox Alpha" model

On OpenRouter the pricing is: Input $0,075/M - Output $0,25/M - Cache Read $0,015 /M

How is the business model of Anthropic/OpenAI will sustain?

Is the actual Z.AI ecosystem good enough to replace the main drivers like Codex and Claude? Because it looks like Z Code is just a Codex fork. Just like the Kimi Code one is.

What irks me about this is that the harnesses seem to be just an afterthought here.

Don't get me wrong, I love messing around with installing Pi, getting it hooked up with OpenRouter, and just trying all kinds of different stuff, local models, etc... but when it comes to literally just setting up a productivity environment and trusting my entire machine with it, I just run Codex.

I have heard from anecdotes where people have indeed replaced their main drivers with DeepSek V4 Flash or GLM and state that "it's almost as good as... [claude/gpt]" but I never hear anyone say "yeah, this is the model/harness that I now run on my machine and don't mess with it"

When reading this type of announcements, always have keen eyes on graphs.

e.g. "Agent Coding Performance by Effort Level" cuts Y-axis from 0~20.

- This makes it as if GLM-5.3-Flash made a bigger jump than it claimed as the Y-axis does not increase much (stupid trick used in biz reports)

I did mention that ox was working ok for me, and having an open-weight comparable to close to SOTA makes it very compelling for me to try it out locally (well, only if I got more VRAM)

It is bench maxed during the stealth testing. And it can’t beat DS flash on speed

At the current 50%-off GLM-5.3-Flash price ($0.075/M input, $0.25/M output; cached input $0.015/M), surprisingly, roughly $400–900/month would buy token throughput comparable to fully exhausting Claude Max 20×

Standard API Pricing for GLM-5.3-Flash (per 1M tokens)

- Input: $0.15 - Output: $0.50 - Cached input: $0.03

Tested this last week and couldn't get it to finish any task that took more then an hour with /goal keep getting errors

Ironically, our administration pushing for ban of the AI chips to China is forcing them to make smaller and more efficient models which seems like a requirement for running on Chinese chips. I wouldn’t be surprised this model was tailored to run purely on Chinese chips. Same thing with Deepseek MLA, the drastically lower KV cache memory requirement was born out of necessity so it runs on the Huawei chips.

Good bicycle, good pelican: https://tools.simonwillison.net/markdown-svg-renderer#url=ht...

It's only 320B, local frontier AI is getting closer, sooner than expected.

Benchmarking is cool, but for production I care about real inference latency, self-hosting VRAM costs, and how cleanly it handles structured JSON output.

When do Chinese models surpass US models? I thought there was at least be a 2 year runway but now I think they surpass it within 12 months, if not sooner.

tbh I wasnt that impressed by it. initial benchmarks were trying to say it was AGI but i told it to re-build Palantir in 1 pass and it gave me a non working prototype

> Combined with our latest 30T-token multimodal pre-training corpus [...]

Is the optimal formula still 20x the amount of model params in tokens for training? Could this mean we're getting a GLM with 1.5t params?

Will we need all the data centers being built or will improvements in software and hardware allow the majority of AI workloads to run locally or in the cloud but way more efficiently than was projected when all the plans were laid out?

Like were executive at Google and AWS and Microsoft expecting this kind of performance from models smaller than what openai/anthropic have been doing? Are we really in a "compute desert"?

Does the word "flash" mean a specific thing when it comes to LLM models? I noticed that this word is used by gemini, qwen, and z.ai and I'm curious does it mean the same thing for each one, or did they all just accidentally brand similarly?

I was quite surprised that Zai had deep pockets to serve this free for a week. My first guess was this was an American lab like xai or google

GLM 5.3 Flash: 320B parameters with 18B activated

Qwen 3.8 Next Flash: 125B + 51B = 176B parameters with 6B activated

DeepSeek V4 Flash: 284B with 13B activated

The new Qwen model is the most promising for one or two Strix Halo 128GB with the low number of active parameters. On paper it's much stronger than Qwen 3.8 27B.

I'm starting to think that this whole sanctioning China may motivate and prompt them to do more and better in every field.

It's too big, bright and resourceful of a country to choose confrontation instead of collaboration.

The key difference between this and all other GLM models is it's multimodal. You cannot send images to the other GLM models.

Related: https://news.ycombinator.com/item?id=49446422

(281 points, 118 comments)

Is anyone actually tried it in agentic coding (claude code loops)? Are apple silicon macs (M5 Max) capable of working with that model? what was the tps?

Luckily, I have the coding plan for z.ai, so I'm happy with this model as I always kept running out of usage with the original glm-5.3

Artificial Analysis benchmark is out: https://news.ycombinator.com/item?id=49450353

So what is a good coding model to run on a 128gb m5 max MacBook nowadays

Has anyone ever asked themselves why AI was made publically available in the first place? is it really economics or is it about training people to recognize the patterns of machine generated words and ideas?

i wonder if more companies will now stealth launch their models. imagine they just released this on openrouter for free but under their normal name - would they get the records in token usage then?

How much is the “discounted” pricing they mention?

In my brief testing, it did about as well as Qwen3.8-4B-Distill, and LFM2.5-2.6B overtook both.

offtopic: Is there any chance we could see competing models from other countries in the next 5 years?

from the article, pareto frontier for open source models is completely dominated by GLM now.

Between Gemma 31/26/12/4/2, Deepseek-v4-flash-0731, Qwen 3.8 27B, Qwen 3.8 Flash Next (which I haven't even gotten to run yet!), and now GLM 5.3 Flash, I can't keep up. I love all these open weight models and am continually stunned that it's largely the West fighting for closed, restrictive, anti-user bullshit and China absolutely mogging the likes of OpenAI and Anthropic, with some notable exceptions like Gemma. Still, I shudder to think what the world would look like if we only had closed models. In many ways the stagnation of open source diffusion seems like that: LLMs are just a few months behind frontier, but image gen is like 1.5 years behind.

Is it good compare to Opus 5 ?

This was the ox-alpha model, right? I remember it performed really well for a model that had 'flash' in its name.

Why is their own coding plan always the last place z.ai release their models? Its even online, you just have to guess the model settings.

> To overcome the relatively limited compute and memory capacity of individual chips, we built a dedicated inference engine for this architecture on top of SGLang. Notably, this effort was accelerated by our GLM-5.3-powered infrastructure agent, which assisted engineers in developing and optimizing kernels, diagnosing performance bottlenecks, and improving the serving stack — creating a feedback loop in which the model helped optimize the system serving the model itself.

> (...) Compared with our initial baseline on the same hardware, we achieved a 3× improvement in end-to-end serving performance, reaching hardware efficiency and per-token cost comparable to mainstream NVIDIA GPUs. This demonstrates that Chinese chips can support frontier-model inference efficiently and economically at scale.

It might be one of the most actually practical tasks that AI might've done because the compounding effects of it and also its implications are/feels so immense. It feels as if Nvidia might be in a slight turbulence from it.

I heard that Dario Amodei is not having a great day today.

2 really strong open models on the same day is a amazing.

so is it confirmed if this is the mysterious OxAlpha model?

This looks like it goes hard, can't wait to try it

It is totally fine I think

By clicking this link you download some PDF in the background

It looks like gemini 3.7 flash actually beats it in a lot of benchmarks, no?

https://x.com/Zai_org/status/2092616204787626030/photo/1

Any providers hosting it outside of China?

Anthropic is accelerating their IPO cause they know what's coming in the next 5 years.

I didn't accept a single edit from this model over the entire week, just saying. I do not understand how it's being benchmarked on par with Sol and other larger models.

[dead]

[flagged]

[dead]

[dead]

[dead]