Zhipu AI logo

GLM-4.5-Air

Zhipu AI

GLM-4.5 Air is a lightweight variant of the GLM-4.5 model by Zhipu AI, optimized for speed and cost-efficiency while retaining strong performance in reasoning, conversation, and general-purpose tasks.

Key Specifications

Parameters
106.0B
Context
-
Release Date
July 28, 2025
Average Score
57.5%

Timeline

Key dates in the model's history
Announcement
July 28, 2025
Last Update
September 15, 2025
Today
November 13, 2026

Technical Specifications

Parameters
106.0B
Training Tokens
-
Knowledge Cutoff
-
Family
-
Capabilities
MultimodalZeroEval

Benchmark Results

Model performance metrics across various tests and benchmarks

Programming

Programming skills tests
SWE-bench-Verified
OpenHands v0.34.0 • Self-reported
57.6%

Reasoning

Logical reasoning and analysis
GPQA
Avg@8 • Self-reported
75.0%

Other Tests

Specialized benchmarks
MMLU-Pro
standard • Self-reported
81.4%
AIME24
Avg@32 • Self-reported
89.4%
MATH-500
standard • Self-reported
98.1%
SciCode
standard • Self-reported
37.3%
LiveCodeBench
2407-2501 • Self-reported
70.7%
TAU-bench-Retail
optimized user simulator • Self-reported
77.9%
BFCL-v3
Full • Self-reported
76.4%
TAU-bench-Airline
optimized user simulator • Self-reported
60.8%
BrowseComp
standard • Self-reported
21.3%
HLE
text-based questions only • Self-reported
10.6%
AA-Index
Estimated • Self-reported
64.8%
Terminal-Bench
Terminus framework • Self-reported
30.0%
Humanity's Last Exam
text-based questions only • Self-reported
10.6%

License & Metadata

License
mit
Announcement Date
July 28, 2025
Last Updated
September 15, 2025

Similar Models

All Models

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.