Skip to main content
Language modelOpen source

GLM-5.3-Flash

Z.ai

Released
August 26, 2026
Data date
September 8, 2026
Benchmark-Meta-Score
Unavailable
Minimum coverage not met
Speed
51.5 tokens/s
Output throughput
Price
$0.15 · $0.5
Input · Output / 1M tokens
Input
Text, Image, Video
Output
Text

GLM-5.3-Flash combines openly available weights with a natively multimodal API. Z.ai lists 320B total parameters, 18B active parameters, a 1M-token context, and text, image, video, and file inputs. The checkpoint is aimed at coding and office work where code, interfaces, and documents meet.

Z.ai describes sparse and linear attention with mHC. Reasoning cannot be disabled in the API; only low, high, and max effort levels are available.

Specifications and access

SpecificationValue and source
Weights
Open weightsSource
Access
API, LocalSource
Parameters
320BSource
Active parameters
18BSource
Context window
1,000,000 tokensSource
Notes
Natively multimodal open-weights MoE with 320B total parameters, 18B active parameters, and a 1M-token context window.Source
License
MITSource
Architecture
hybrid sparse + linear attention with mHC; natively multimodalSource
Parameters
320 BSource
Active parameters
18 BSource
Input
Text, Image, VideoSource
Output
TextSource
Tool calling
YesSource
Training tokens
30T multimodal pretraining tokensSource
API model ID
glm-5.3-flash (API-model identifier is supported by the read card identity; the separate API guide was not successfully opened.)Source
Reasoning mode
Thinking and non-thinking modes; reasoning effort: low, high, max (default max).Source