GLM-5.5 Release Date: Not Announced Yet. The Rumors, Checked

GLM-5.5 isn't released yet. Where the name came from, which rumors were denied, what GLM-5.3 costs today, and three commands that show when it ships.

AIHubMix6 min czytania
GLM-5.5 Release Date: Not Announced Yet. The Rumors, Checked

GLM-5.5 has not been released. As of November 19, 2026 there is no GLM-5.5 model page, API model ID, price, or open-weight repository, and Z.ai's own pages don't use the name. The newest models from Z.ai (Zhipu AI) are still the GLM-5.3 family. The earliest mention found is one sentence in a June news story that cites no source of its own. August brought GLM-5.3 instead, the only next-generation name Zhipu has given is GLM-6.0, and the November leak about GLM-5.5's sizes was publicly denied by a Z.ai researcher.

Below: where the name came from, what is confirmed, what to use now, and three commands that show whether a successor has shipped.

Is GLM-5.5 out? Five places to check

Every listing below was checked on November 19, 2026, between 07:00 and 08:00 UTC.

Where Newest GLM listed GLM-5.5
Z.ai release notes GLM-5.3-Flash, Aug 26 Not listed
Z.ai pricing page GLM-5.3 family No price
Hugging Face, zai-org GLM-5.3 and 5.3-Flash No public repo
OpenRouter GLM-5.3-FlashX-Legal, Nov 2 No model ID
AIHubMix model list GLM-5.3 family Page not found

Z.ai's release notes end at GLM-5.3-Flash on August 26, and the documentation paths docs.z.ai/guides/llm/glm-5.4 and glm-5.5 return 404. Zhipu's Chinese release notes stop at the same entry.

Where the GLM-5.5 name came from

The earliest mention is a news report, not a Z.ai announcement. Here is how the story grew, next to what Z.ai actually shipped (the news reports are linked under Sources):

  • June 30: a CGTN story on the GLM-5.2 launch ends with "The company's next model, GLM-5.5, is expected in August." The sentence names no source of its own. It closes a paragraph that opens "Reuters also reported", and the JP Morgan forecast quoted there is about revenue, not models.
  • July: posts on X added detail: Z.ai would skip GLM-5.3 and ship GLM-5.5 in August with over a trillion parameters, open weights, and a 1M context. A later post raised the size past 3 trillion.
  • August: Z.ai shipped GLM-5.3 (API on August 18) and GLM-5.3-Flash (August 26). Flash had been tested anonymously as "Ox Alpha" before launch.
  • August 31: at Zhipu's half-year results call, founder Tang Jie described the next-generation model as GLM-6.0, a "Full Self-Training" model. Guandian's report of the call gives no release date and does not mention GLM-5.5.
  • September 4: a new anonymous model, "Omen Alpha", appeared on OpenCode, and BlockBeats reported that OpenCode's data page listed it under Zhipu. Zhipu hadn't claimed it then, and Z.ai's docs still don't mention it.
  • November 4: an X post said Tang Jie had revealed "the upcoming GLM-5.5 (32B, 122B, and 256B)" in a Tsinghua lecture. No recording or event page has surfaced.
  • November 8: Cunxiang Wang, whose X profile lists him as a Z.ai researcher, replied to a follow-up post from the same account that the "internal 32B, 122B, and 256B checkpoints" were "fake news". His profile also says his opinions are his own, so this is not an official statement.

What's confirmed, what's rumor

Claim Status
A model named GLM-5.5 Not confirmed
GLM-5.5 in August Did not happen
Over 1T (later 3T) parameters Rumor
32B, 122B and 256B sizes Denied
GLM-6.0 is next-generation Stated, no date
GLM-5.3 is the flagship Confirmed

"Denied" means a Z.ai researcher called the sizes fake on his personal account. A later post that relayed the denial guessed at two models, GLM-5.5 and a Flash version, with about 1.6T parameters; it cited nothing, so treat it as another rumor.

Z.ai has shipped a numbered GLM-5 release every 54 to 70 days this year, and November 19 is 93 days after GLM-5.3. That pattern makes a release soon plausible but proves nothing, including its name: the next model could be called GLM-5.4, GLM-5.5, or GLM-6.0, or appear first under a codename the way Flash did.

What to use until it ships

GLM-5.3 is the current flagship. Z.ai's GLM-5.3 guide says it uses the same base model as GLM-5.2, with every gain coming from post-training. These are the three models on Z.ai's latest-models price list, at its official prices per million tokens:

Model Input Cached input Output Context
GLM-5.3 $1.40 $0.26 $4.40 1M
GLM-5.3-Flash $0.15 $0.03 $0.50 1M
GLM-5.3-FlashX $0.37 $0.075 $1.25 1M

Prices are from Z.ai's pricing page; cache storage is listed as free for a limited time. All three output up to 128K tokens.

  • GLM-5.3 takes text only, with thinking always on: reasoning_effort accepts low, high, or max, and defaults to max. Its weights ship under Z.ai's own GLM-5.3 License, not MIT.
  • GLM-5.3-Flash has 320B total parameters with 18B active, accepts images and video as well as text, and is MIT-licensed.
  • GLM-5.3-FlashX is a faster version of Flash; Z.ai quotes inference speeds of 200 tokens per second. AIHubMix hasn't yet verified its thinking, vision or tool support, so test before relying on them.

All three are on AIHubMix under the IDs glm-5.3, glm-5.3-flash, and glm-5.3-flashx. The model guides list the Chat Completions endpoint (POST /v1/chat/completions):

import os
from openai import OpenAI

client = OpenAI(base_url="https://aihubmix.com/v1", api_key=os.environ["AIHUBMIX_API_KEY"])

resp = client.chat.completions.create(
    model="glm-5.3",
    reasoning_effort="high",  # low, high, or max (default)
    messages=[{"role": "user", "content": "Review this function for off-by-one errors: ..."}],
)
print(resp.choices[0].message.content)

When a successor ships, switching may take only a new model ID. Check its thinking parameters first, though: the move from GLM-5.2 to GLM-5.3 removed the option to turn thinking off.

How to tell when it ships

The first two commands print the newest entries in Z.ai's release notes and on Hugging Face; the third prints any GLM model ID on AIHubMix newer than 5.3, including IDs with a prefix such as coding-:

# Z.ai release notes: newest dated entries
curl -s https://docs.z.ai/release-notes/new-released.md \
  | grep -A2 '^\*\*20' | grep -vE '^[[:space:]]*$|^--$' | head -6

# Hugging Face: newest repositories under zai-org
curl -s 'https://huggingface.co/api/models?author=zai-org&sort=createdAt&direction=-1&limit=5' \
  | grep -o '"id":"[^"]*"'

# AIHubMix: any GLM model newer than 5.3
curl -s 'https://aihubmix.com/api/v1/models?type=llm&sort_by=order' \
  | grep -oE '"model_id":"[^"]*glm-(5\.[4-9]|[6-9])[^"]*"' || echo "nothing newer than GLM-5.3 yet"

Weights can trail the API: GLM-5.3's weights reached Hugging Face around August 28, ten days after its August 18 API launch.

To skip the polling, turn on the "New model" alert in the AIHubMix console. It covers every model and can notify you on the web, by email, or through a webhook.

Treat an anonymous model on OpenCode or OpenRouter as a hint, not a launch: Zhipu tested Flash under a codename, and its real name arrived only with the release notes.

FAQ

Is GLM-5.5 available through an API?
No. Z.ai, OpenRouter and AIHubMix list no GLM-5.5 model ID, and Z.ai hasn't announced one. A page selling "GLM-5.5 access" today isn't selling an official Z.ai model.

When will GLM-5.5 be released?
Zhipu hasn't given a date. The August 2026 date in news reports passed, and August brought GLM-5.3 instead.

Will the next GLM be called GLM-5.5?
Unknown. Zhipu's only stated next-generation name is GLM-6.0, and Z.ai's own pages don't use the name GLM-5.5.

How big will GLM-5.5 be?
No reliable figure exists. Rumors range from 32B to over 3 trillion parameters, and a Z.ai researcher called the 32B, 122B and 256B sizes fake.

Is Omen Alpha GLM-5.5?
Unconfirmed. BlockBeats reported in September that OpenCode's data page listed it under Zhipu, but Z.ai's docs don't mention it.

What should I use until then?
GLM-5.3 for coding and agent work, or GLM-5.3-Flash when you need image or video input at a lower price.

Will GLM-5.5 be open-weight?
Unknown. Recent GLM releases published weights on Hugging Face, though GLM-5.3 uses a custom license rather than MIT.

Sources

  1. Z.ai: release notes, pricing, and the GLM-5.3 and GLM-5.3-Flash guides.
  2. CGTN: Chinese AI steps onto global stage as GLM-5.2 narrows frontier gap, June 30, 2026.
  3. Guandian: Tang Jie on GLM-6.0, August 31, 2026, and BlockBeats on Omen Alpha, September 4, 2026.
  4. CellCog: GLM-5.5 release date: the leak vs what Z.ai shipped, which compiles the July posts on X.
  5. AIHubMix: GLM-5.3 model page. Listings and X posts checked November 19, 2026, 07:00–08:00 UTC.

Jedna brama do wielu dostawców AI. Stworzona z myślą o produkcji.

Odwiedź AIHubMix