Prototype · closed beta
unOpen.AI
In closed beta: Roxy, a private AI assistant with permanent memory running on infrastructure the user owns. Brand, site, product, backend and conversational design.
Model families
Z.ai GLM.
Open weights, read the licence.
Z.ai, formerly Zhipu AI, ships GLM both as downloadable weights and as API-only tiers. That makes it a serious candidate for self-hosted coding and agent work, provided the licence on the exact release you deploy allows it.
All Model Families →
Current release
Z.ai GLM today.
Z.ai ships these models as a range rather than one lead model, so this page is about the family: what it is good at, where it runs and how we evaluate it.
Last reviewed on 30 October 2026. We check the public model catalogues every week; when a new release appears, we review it against the provider's announcement and update this page.
LATEST MODELS
The newest Z.ai GLM models.
What Z.ai has shipped most recently. Each has its own page with specifications, independent ratings at every reasoning setting and listed prices.
23 Sep 2026
GLM 5.3 Prime
- Intelligence
- —
- Context
- 1M
- In / out per M
- $2.80 / $8.80
18 Sep 2026
GLM 5.3 FlashX
- Intelligence
- —
- Context
- 1.05M
- In / out per M
- $0.370 / $1.25
26 Aug 2026 · open weights
GLM 5.3 Flash
- Intelligence
- 27.8
- Context
- 1.05M
- In / out per M
- $0.018 / $0.140
18 Aug 2026 · open weights
GLM 5.3
- Intelligence
- 31.7
- Context
- 1.05M
- In / out per M
- $0.080 / $3.60
EVERY MODEL
All 16 Z.ai GLM models, compared.
Sort by any column. Select a model for its full page.
| GLM 5.3 Prime | 23 Sep 2026 | — | — | 1M | $2.80 | $8.80 |
|---|---|---|---|---|---|---|
| GLM 5.3 FlashX | 18 Sep 2026 | — | — | 1.05M | $0.370 | $1.25 |
| GLM 5.3 Flash open | 26 Aug 2026 | 27.8 | 66 | 1.05M | $0.018 | $0.140 |
| GLM 5.3 open | 18 Aug 2026 | 31.7 | 76 | 1.05M | $0.080 | $3.60 |
| GLM 5.2 open | 16 Jun 2026 | — | — | 1.05M | $0.006 | $7.00 |
| GLM 5.1 open | 7 Apr 2026 | — | — | 205k | $1.40 | $4.40 |
| GLM 5V Turbo | 1 Apr 2026 | — | — | 203k | $1.20 | $4.00 |
| GLM 5 Turbo | 15 Mar 2026 | — | — | 203k | $1.20 | $4.00 |
| GLM 5 open | 11 Feb 2026 | — | — | 205k | $1.00 | $3.20 |
| GLM 4.7 Flash open | 19 Jan 2026 | — | — | 200k | $0.061 | $0.400 |
| GLM 4.7 open | 22 Dec 2025 | — | — | 205k | $0.540 | $1.98 |
| GLM 4.6V open | 8 Dec 2025 | — | — | 131k | $0.300 | $0.900 |
| GLM 4.6 open | 30 Sep 2025 | — | — | 205k | $0.500 | $2.00 |
| GLM 4.5V open | 11 Aug 2025 | — | — | 66k | $0.600 | $1.80 |
| GLM 4.5 open | 28 Jul 2025 | — | — | 131k | $0.600 | $2.20 |
| GLM 4.5 Air open | 28 Jul 2025 | — | — | 131k | $0.130 | $0.850 |
Intelligence and coding: Artificial Analysis indices at the model's best reasoning setting, and its measured output speed. Context and prices: the OpenRouter catalogue list price in USD per million tokens. Released: the provider's date where Artificial Analysis records one, otherwise the date the catalogue first listed it. Refreshed weekly; not measured by QQuantum.ai. Select a heading to sort.
Specification
- Provider
- Z.ai (formerly Zhipu AI), Beijing, China
- Weights
- Published for much of the range on Hugging Face (zai-org); some tiers, such as GLM 5.3 Prime and FlashX, are offered by API only
- Licence
- Differs by release: GLM-5.3 carries Z.ai's own GLM-5.3 licence, while GLM-5.3-Flash is MIT. Read the model card of the exact release you deploy
- Access paths
- Z.ai's API, third-party model catalogues, or an engineered deployment of the published weights
- Where we would test it
- Self-hosted coding and agent work, and as an open-weight alternative in a routed model pool
Provider-published specifications as of 30 October 2026, not figures we measured. Model families move quickly — check the provider's own documentation before committing to a number.
Licences: the GLM-5.3 and GLM-5.3-Flash model cards, checked 30 October 2026.
INDEPENDENT RATINGS
How Z.ai GLM compares.
Third-party scores and listed prices, refreshed every week. Useful for building a shortlist; your own evaluation decides.
Composite index across their reasoning, knowledge, maths and coding evaluations · Higher is better · Linear scale from 0 to 66.0
- Gemini 4 Carbon
- 59.4
- Claude Opus 5.5
- 58.4
- GPT-6 Astra
- 56.9
- Muse Wonder Full
- 49.7
- GLM-5.3
- 31.7
- GLM-5.3-Flash
- 27.8
This family's highest-rated models, highlighted, against the best-rated model of each other family we cover. One composite score cannot show cost, latency, data residency or how a model handles your task.
Source: Artificial Analysis. Checked 30 October 2026. Measured independently by Artificial Analysis, not by QQuantum.ai or the providers.
USD per million output tokens · The newest releases in the family · Linear scale from 0 to $10.00
- GLM 5.3 Prime
- $8.80
- GLM 5.3 FlashX
- $1.25
- GLM 5.3 Flash
- $0.14
- GLM 5.3
- $3.60
- GLM 5.2
- $7.00
Input is priced separately: GLM 5.3 Prime $2.80, GLM 5.3 FlashX $0.37, GLM 5.3 Flash $0.02, GLM 5.3 $0.08, GLM 5.2 $0.01 per million tokens. Catalogue list prices; the provider's own rates, caching and batch discounts can differ.
Source: OpenRouter model catalogue. Checked 30 October 2026. Listed by OpenRouter, not measured by QQuantum.ai.
Where we would choose something else
Where the organisation has a standing policy against models from Chinese providers, that policy decides, and Mistral or Meta's models cover adjacent ground. Where it does not, the useful distinction is between using Z.ai's hosted API, which sends data to the provider, and running published weights on hardware you control, which does not.
Self-hosting the largest GLM releases is an infrastructure project: size the weights, context cache and concurrency together before buying hardware. A smaller open tier frequently does the narrow task at a fraction of the cost, and a replay evaluation on your own requests settles which.
Questions
Questions about Z.ai GLM.
Is GLM open source?
Many GLM releases publish their weights, but the licence differs by release: some are MIT, others carry Z.ai's own licence with its own conditions. Treat each release separately and keep the licence with the deployed artefact.
Does using GLM send our data to China?
Using Z.ai's hosted API sends requests to the provider, so review its processing locations and terms. Running published weights on your own hardware does not transmit anything by itself; inspect the serving stack and any tools for outbound connections as you would for any self-hosted model.
How do we decide between GLM and another open model?
Run the same representative tasks on two or three candidates and compare accepted results, latency and total cost. The tables above give a shortlist; your own evaluation decides.
