GLM-5.3-Flash
Open-weight multimodal intelligence at one-tenth of the flagship price.
Z.ai's budget open-weight model with native image and video input, MIT-licensed weights, and a 1M-token context window.

GLM-5.3-Flash model card on Hugging Face
Overview
What is GLM-5.3-Flash?
Open-weight multimodal intelligence at one-tenth of the flagship price.
GLM-5.3-Flash is the budget model in Z.ai's GLM-5.3 family, released on 26 August 2026. It is a separately trained base model, not a distilled or quantised version of the flagship: 320 billion total parameters with 18 billion active per token, using hybrid sparse and linear attention.
It is the first natively multimodal model in the GLM-5 family, accepting text, image and video input, with the same 1,048,576-token context window as the flagship. Its weights are published on Hugging Face under the permissive MIT licence. On Artificial Analysis' independent Intelligence Index v5.0.1 it scores 28. Z.ai's API lists $0.15 per million input tokens and $0.50 per million output tokens.
Platforms and languages
GLM-5.3-Flash Availability
Platforms
Languages
Capabilities
GLM-5.3-Flash Key Features
320B MoE, 18B active
A separately trained base model with hybrid sparse and linear attention, not a cut-down flagship.
Natively multimodal
Accepts text, image and video input, the first multimodal model in the GLM-5 family.
MIT-licensed weights
Plain MIT licence on Hugging Face, with no extra commercial-use review for hosted serving.
1M-token context window
Same long context as the flagship for documents, code and agent runs.
Low API pricing
Z.ai lists $0.15 per million input tokens and $0.50 per million output tokens.
Plans and access
GLM-5.3-Flash Pricing
Paid
Common questions
GLM-5.3-Flash FAQs
How much does GLM-5.3-Flash cost?
Z.ai lists $0.15 per million input tokens, $0.03 cached, and $0.50 per million output tokens.
Is GLM-5.3-Flash a distilled version of GLM-5.3?
No. Z.ai says it is a newly trained base model with its own architecture and training recipe.
What licence does GLM-5.3-Flash use?
MIT. The weights on Hugging Face carry a standard permissive MIT licence.
Can GLM-5.3-Flash handle images?
Yes. It accepts text, image and video input natively.
Reviews
See what the community thinks and share your experience.
—
Based on 0 ratings
Rating distribution
Community reviews
No reviews
No written reviews yet.