GLM 5.3: What Z.ai Shipped, and the Weights It Held Back

Update, October 7, 2026: the weights did arrive on schedule. Z.ai published the GLM 5.3 weights on Hugging Face on August 28, 2026, exactly two weeks after launch, after what Z.ai lead Zixuan Li called "two additional weeks of comprehensive safety evaluations". They ship under a new GLM-5.3 License rather than MIT. It grants the same broad rights, with one condition: a company that sells GLM 5.3 as a model service and has more than $10 billion in revenue over 12 months must pass Z.ai's security review before commercial use. The sections below have been updated where they described the weights as missing.

Update, August 26, 2026: the roughly two-week timeline described below has now elapsed, and GLM 5.3’s open weights are still not published. Z.ai instead released GLM 5.3 Flash, a separate 320B multimodal model on a newly trained base, with MIT-licensed weights on day one. The 744B GLM 5.3 weights described in this article remain unreleased.

Z.ai released GLM 5.3 on August 14, 2026, and the launch is unusual in one very specific way. This is not a new architecture and not a new pretrain. GLM 5.3 runs on the same 744-billion-parameter base as its predecessor, and every reported gain comes from extended post-training. By the company’s own figures it now leads CyberGym at 84.5 percent, ahead of both Claude Mythos 5 and GPT-5.6 Sol.

It also shipped without the open weights that made the GLM line matter in the first place. Z.ai said API access and weights would arrive in stages after a safety evaluation, roughly two weeks out, and the weights followed on August 28. That decision is the real story here. Below is what actually changed, what the benchmark chart quietly leaves out, why Z.ai says it held the weights back, and how to reach GLM 5.3 in the meantime.

The Key Takeaways

• Released: August 14, 2026, post-trained on the same 744B base as GLM 5.2. No new pretrain.
• Biggest jump: Terminal-Bench 3.0 went from 4.6 to 28.3, and DeepSWE v1.1 from 46.2 to 66.9.
• Security lead: CyberGym rose to 84.5 percent, edging Mythos 5 at 83.8 and GPT-5.6 Sol at 83.6, according to Z.ai’s own chart.
• Weights held back, then released: a break from GLM 5.2, which shipped MIT-licensed weights within days. Z.ai held GLM 5.3 for a safety review and published the weights on August 28, 2026, under a custom licence instead of MIT.
• Where to get it: the GLM Coding Plan from $18 a month, ZCode, and the public API at $1.40 in / $4.40 out per million tokens, the same rate as GLM 5.2.

What GLM 5.3 Actually Is

GLM 5.3 is Z.ai’s coding and agentic model, and the headline technical fact is what did not change. The base model is identical to the one underneath GLM 5.2 and its 1M-token context. Z.ai did not retrain it. Instead the company scaled post-training: more task environments, more environment types, and longer training runs on top of the existing 744B foundation.

That matters more than it sounds. For most of the past two years, a version bump in a frontier model line meant a bigger or freshly trained base. GLM 5.3 shows a lab pulling a large capability jump out of post-training alone, on hardware it has already paid for. New to this model family? The background on what GLM actually is covers how Zhipu’s open line got here.

Z.ai framed the launch on X in three lines, and the framing is worth reading closely, because “cyber defense” is doing a lot of work in it.

GLM 5.3 Benchmarks: the Wins and the Losses

Every number below comes from Z.ai’s own launch chart. There is no independent audit of GLM 5.3 yet, and the comparison figures for rival models are the ones Z.ai chose to publish.

Read them as a vendor’s best case, not a neutral scoreboard.

Where GLM 5.3 leads

The Terminal-Bench result is the one to look at.

Going from 4.6 to 28.3 is not an incremental gain, it is a model that could barely hold a long terminal session becoming one that can. DeepSWE v1.1 climbing twenty points tells a similar story about sustained, multi-step work rather than single-turn answers.

On security, CyberGym at 84.5 percent puts GLM 5.3 fractionally ahead of the closed frontier models Z.ai lists. That is a genuine milestone for an open-weight line, and since August 28 the weights are downloadable too.

Where it still trails

The chart is less flattering elsewhere, and most launch coverage skipped this part.

On Z.ai’s own internal Code Bench, GLM 5.3 reaches 31.4 percent at a 50,000-token budget against Claude Opus 4.8 at 29.5 percent, but Claude Fable 5 leads the same test at 39.5 percent. On ExploitBench, the jump from 24.4 to 54.4 is dramatic in relative terms and still well behind the frontier leaders, which Z.ai reports at 78.0 percent.

So the honest summary is narrower than the headlines: GLM 5.3 is at or near the frontier on vulnerability discovery and long-horizon agentic coding, and clearly behind it on raw code quality and on exploitation. For a wider view of how this line compares across the board, see how GLM stacks up against Claude, GPT and Gemini.

The Capability That Outgrew Its Training

The most interesting disclosure in this launch is not a benchmark. Z.ai added vulnerability discovery environments to post-training expecting the model to get better at spotting individual bugs. According to the company, it got something else. Capability “continued compounding as training scaled, and the model began reasoning across multiple stages of exploitation, forming coherent plans for complete exploitation chains.”

In plain terms, Z.ai aimed for a better bug finder and produced a model that plans across a whole attack chain. The company also reports real-world output from this line: 2,436 vulnerabilities identified across 269 open-source projects since GLM 5.2, with 1,097 rated critical or high severity. Those are defensive findings, and they are the strongest argument for the “cyber defense” framing.

They are also, unavoidably, evidence of the same capability pointed the other way.

This is not the first time in 2026 that a lab has described exactly this problem. It is worth reading alongside the running record of AI safety incidents in 2026, and the case where AI models attacked Hugging Face.

Why the Open Weights Came Two Weeks Late

GLM 5.2 shipped MIT-licensed weights to Hugging Face within days of launch. That was the whole proposition: frontier-adjacent performance you could download, inspect and self-host. GLM 5.3 broke that pattern on day one. At launch there was no GLM 5.3 repository on Z.ai’s Hugging Face organisation, where the only entries added since are the GLM 5.3 Flash repositories published on August 26, 2026.

Z.ai’s stated reason is the capability described above. The company says API access and open weights will be released in stages following safety evaluations, and reporting from Unite.AI puts that at roughly two weeks, so the end of August 2026.

If that sounds familiar, it should. OpenAI made a structurally identical decision this month, which we covered in OpenAI’s hacking model and why you cannot use it. Two labs, two weeks apart, both concluded that a cyber-capable model needed gating before general release.

That is a pattern, not a coincidence. It has real consequences for anyone whose roadmap assumes downloadable frontier models, and the current state of play is in our guide to the best open-source AI models.

Z.ai kept that promise. The weights landed on Hugging Face on August 28, exactly two weeks after launch. What changed is the licence. GLM 5.2 used plain MIT. GLM 5.3 ships under a GLM-5.3 License that grants the same rights to use, modify, fine-tune and sell, but adds one clause: a Model-as-a-Service business whose revenue tops $10 billion over any 12 months must pass Z.ai's security review before commercial use. For individuals, startups and almost every company, nothing changes. For the largest cloud providers, GLM 5.3 is no longer a free drop-in.

How to Use GLM 5.3 Today

Access is narrow at launch. Z.ai’s follow-up post states that GLM 5.3 is available through the GLM Coding Plan and ZCode, with broader API access staged behind the same safety review. At launch, the public subscribe page still advertised the plan as powered by GLM 5.2 and GLM-5-Turbo. That has since settled: Z.ai's Coding Plan docs now say requests for GLM-5.2 and GLM-5.1 are routed automatically to GLM-5.3.

The Coding Plan runs at three tiers: Lite at $18 a month, Pro at $80 and Max at $168. Annual billing takes 30% off, for effective monthly rates of $12.60, $56 and $117.60, and quarterly billing takes 20% off. Usage is metered in credits rather than prompts: 2,000, 12,000 and 28,000 credits per five hours on Lite, Pro and Max, with weekly caps of 10,000, 60,000 and 140,000.

The per-token rate is now public too. Z.ai's rate card lists GLM 5.3 at $1.40 per million input tokens, $0.26 for cached input and $4.40 per million output tokens, identical to GLM 5.2, so moving API traffic from 5.2 to 5.3 costs nothing extra. Our breakdown of GLM pricing and the Coding Plan tiers covers how that quota system behaves in practice.

Using GLM 5.3 on a Mac

ZCode is Z.ai’s own coding agent, and it is optional rather than mandatory. The Coding Plan itself works with Claude Code, Cline, Roo Code and a long list of other harnesses, which is the path most Mac developers will want. There is no weight download to wait on for any of those routes, because the model runs server-side either way.

There is another option if you would rather not buy a coding subscription just to try one model. Fello AI already puts GLM alongside Claude, GPT, Gemini and Grok in a single native Mac app, which is the practical route if you do not have server hardware for the 5.3 weights. The GLM desktop client for macOS page covers that setup.

Is GLM 5.3 Worth Switching To

If you are already on the GLM Coding Plan, this is a free upgrade to a meaningfully better agentic coder, and the Terminal-Bench and DeepSWE jumps are large enough to notice in daily use.

Take it.

If you were waiting on GLM 5.3 specifically to self-host a frontier-class open model, the wait is over, as long as the licence fits you. Xiaomi published MIT weights for MiMo-V2.6 on September 21, 2026, and Artificial Analysis now scores it above GLM 5.3, so there is a downloadable alternative today that did not exist in August. The value of this line has always been the licence as much as the scores, and GLM 5.3's is slightly stricter than GLM 5.2's MIT. Tencent took the opposite route on the same day, publishing Hy4 Preview’s 770B weights under plain Apache 2.0.

Z.ai now offers two downloadable 5.3 models. GLM 5.3 Flash, a different and much smaller model, arrived two days earlier under plain MIT, and the full GLM 5.3 followed under its own licence. Our notes on what is rumoured for GLM 5.5 track what comes after.

And if you are comparing coding models on merit rather than licence, GLM 5.3 has not displaced Claude at the top of the code-quality tests, by Z.ai’s own admission. The verdict for now: an excellent agentic coder, a genuine security milestone, and an open model that is open again, on slightly stricter terms.

FAQ