GLM 5.5 is reportedly expected in August 2026, according to a June 30 report from CGTN on Zhipu AI. That would make it the direct successor to GLM-5.2, the open-weight model that shipped on June 13 and immediately became the highest-scoring open-source model on the Artificial Analysis Intelligence Index with a score of 51. Zhipu has not published an official launch date, so August is a strong signal rather than a locked promise.
This article pulls together every credible signal on GLM 5.5 in one place. You will get the reported timing, the release cadence that backs it up, a clear-eyed view of what the specs and price are likely to look like, and where it fits in Zhipu’s bigger plan to ship a frontier-class open model before the end of the year. We flag what is confirmed and what is still speculation, so you know exactly how much weight to put on each claim.
The Key Takeaways
- GLM 5.5 is reportedly expected in August 2026, per a June 30 CGTN report, as the successor to GLM-5.2.
- Nothing about GLM 5.5 is officially confirmed yet, no date, no specs, no price from Zhipu itself.
- The two-month release cadence (GLM-5 in February, GLM-5.1 in April, GLM-5.2 in June) lines up neatly with an August drop.
- Expect GLM 5.5 to stay open weights under the MIT license, like GLM-5.2, at roughly a sixth of frontier API cost.
- Zhipu has publicly committed to a Fable/Mythos-class open model by the end of 2026, and GLM 5.5 is the likely next step toward it.
Is GLM 5.5 Officially Confirmed?
No. As of late June 2026, Zhipu AI (which now markets itself as Z.ai) has not made any official GLM 5.5 announcement. The only public timing comes from a CGTN report covering the GLM-5.2 launch, which states that the company’s next model, GLM 5.5, is expected in August.
That single line is the entire confirmed basis for the date. There is no spec sheet, no benchmark, no pricing and no exact day attached to it yet. Treat everything beyond “expected in August” as informed extrapolation from Zhipu’s recent track record.
Keep the version naming in mind too. Jumping from 5.2 straight to 5.5 skips 5.3 and 5.4, which suggests either a larger-than-usual upgrade or simply Zhipu’s habit of moving fast and renumbering on its own logic. Until the company confirms it, even the 5.5 label is provisional.
Why an August Release Date Is Plausible
The cadence is the strongest argument. Zhipu has shipped a new GLM 5.x flagship roughly every two months through 2026, and an August release sits exactly on that rhythm. Each version has landed as downloadable open weights with quick follow-up benchmarks, so the pattern is well established.
Here is how the recent GLM 5.x line has rolled out.
| Model | Released | Params (MoE) | Status |
|---|---|---|---|
| GLM-5 | Feb 13, 2026 | 744B total / 44B active | Superseded |
| GLM-5.1 | Apr 2026 | 744B total | Superseded |
| GLM-5.2 | Jun 13, 2026 | 744B total / 40B active | Current flagship |
| GLM 5.5 | ~Aug 2026 (expected) | Not announced | Rumored |
Two months between each release puts the next one in August, which matches the CGTN line precisely. The company also has obvious commercial reasons to keep the pace up. GLM-5.2 sent Zhipu’s stock higher and earned headlines for beating GPT-5.5 on agentic coding at a fraction of the cost, so momentum is firmly on its side.
What to Expect from GLM 5.5
Everything in this section is extrapolation from the GLM 5.x trajectory, not confirmed fact. With that caveat front and center, here is the most likely shape of the release based on how GLM-5.2 looks today.
| Area | GLM-5.2 (current) | GLM 5.5 (expected) |
|---|---|---|
| Release | Jun 13, 2026 | ~August 2026, unconfirmed |
| License | MIT open weights | Expected MIT open weights |
| Context window | 1 million tokens | 1M tokens or larger |
| Benchmark target | 51 on Intelligence Index, top open-source | Closing the last 1-4 points to Claude Opus 4.8 |
| API cost | ~$1.40 in / $4.40 out per M tokens | Expected similar, ~1/6 of frontier pricing |
| Focus | Agentic coding, long-horizon tasks | More coding and agentic gains, likely multimodal push |
Benchmarks and capability
GLM-5.2 already ranks first among globally available models on Code Arena and trails Claude Opus 4.8 by just 1 percentage point on FrontierSWE and 4 points on Terminal-Bench 2.1. GLM 5.5’s obvious job is to close that last gap. Expect Zhipu to push hard on agentic coding and long-horizon task performance, since that is where the model already competes and where the headlines come from.
Context, parameters and efficiency
The 1-million-token context window arrived with GLM-5.2, so GLM 5.5 is more likely to refine efficiency than chase a bigger headline number. The current model is a 744-billion-parameter mixture-of-experts design that only activates around 40 billion parameters per token, which is how Zhipu keeps inference cheap. Expect that efficient MoE approach to carry forward, possibly with smaller Air or Flash variants for consumer hardware.
Open weights and pricing
Every recent GLM flagship has shipped under the permissive Licence MIT with no regional locks, and GLM-5.2 supports several domestic Chinese chip architectures out of the box. There is no reason to expect GLM 5.5 to break that pattern. The free downloadable weights will likely sit alongside a paid hosted API and the GLM Coding Plan, at the same aggressive pricing that has made the family popular.
The Bigger Picture: Road to “Open Fable”
GLM 5.5 is not the end goal. Zhipu has publicly committed to releasing an open-weight model at Fable/Mythos level, the current frontier tier, before the end of 2026. In the AI community this target has picked up the nickname “Open Fable,” and it frames GLM 5.5 as a stepping stone rather than the destination.
The ambition is not subtle. In a brief public exchange noted by Latent Space’s AINews, Elon Musk suggested GLM could reach Mythos-level on benchmarks by year-end, and Zhipu CEO Tang Jie replied that it “won’t take that long.” If GLM 5.5 lands in August and a year-end flagship follows, Zhipu would be on track to put genuinely frontier-class capability into fully open weights, something no Western lab currently offers at this license or price.
That is the real reason GLM 5.5 matters beyond the version number. It is the next data point on whether open-weight Chinese models can fully close the gap with the best AI models from OpenAI, Anthropic and Google.
How to Try GLM Without the Wait
You do not need a Z.ai account or a local GPU rig to see what the fuss is about. Inside Fello AI you can swipe between frontier models in one chat, run the same prompt across several at once, and compare answers side by side. When a new flagship like GLM 5.5 lands, you get to use it without setting up an API key or hosting anything yourself.
That side-by-side setup is the fastest way to judge a release on your own work rather than on benchmark charts. Drop a real coding task or a long document into GLM-5.2 today, then re-run it against GLM 5.5 the day it drops, and you will see the actual difference for your use case.
Závěr
GLM 5.5 is shaping up to be Zhipu’s next move in a remarkably fast release cycle, with an August 2026 window that fits both the reported timing and the two-month cadence. Until the company confirms it, treat the date, specs and price as well-grounded expectations rather than facts. The smart play is to get familiar with GLM-5.2 now, since GLM 5.5 will almost certainly build on the same open-weight, low-cost foundation.
If you want to be ready, read up on what GLM is and the current GLM 5.2 flagship, then keep this page bookmarked. We will update it the moment GLM 5.5 goes live.
FAQ
When is GLM 5.5 coming out?
GLM 5.5 is reportedly expected in August 2026, based on a June 30 CGTN report about Zhipu AI. It would be the direct successor to GLM-5.2, which launched on June 13. Zhipu has not confirmed an exact date, so treat August as a strong signal rather than a locked launch.
Is GLM 5.5 confirmed by Zhipu?
No. There is no official Zhipu or Z.ai announcement for GLM 5.5 yet. The only public timing comes from a news report, and there are no official specs, benchmarks or pricing for the model so far.
Will GLM 5.5 be free and open source?
Almost certainly. Every recent GLM flagship, including GLM-5.2, shipped as downloadable open weights under the permissive MIT license. Expect GLM 5.5 to follow the same open-source playbook, with a paid hosted API and Coding Plan alongside the free weights.
How will GLM 5.5 compare to GLM-5.2?
Expect incremental gains rather than a total redesign. GLM-5.2 already trails Claude Opus 4.8 by just 1 to 4 points on key coding benchmarks, so GLM 5.5’s likely goal is to close that gap while keeping the 1-million-token context window and low cost.
What is “Open Fable”?
It is the community nickname for Zhipu’s stated goal of releasing an open-weight model at frontier Fable/Mythos level before the end of 2026. GLM 5.5 is widely seen as a stepping stone toward that year-end target.
