#99 GLM 5.3
high- Cost
- $0.085
- Time
- 362.5s
- Tokens
- 19,310 tok
Z.ai
Release: 2026-08-20
Tested on: 2026-08-20 14:09
z-ai/glm-5.3::high
744B total (40B active)MoEClosed
(high)
(low)
Summary
GLM 5.3 scores 7.3 on AI BENCHY and ranks #99. It has 10.0 reliability, a 74.2% pass rate, $0.403 total cost, and 27.82s average response time.
Model facts
Researched on 2026-08-20
- Parameters
- 744B total (40B active)
- Architecture
- MoE
- Availability
- Closed
- License
- -
Z.ai reports that GLM-5.3 uses the same base model as GLM-5.2. The announced weights were not public at research time.
7.3
Consistency
7.5
10.0
$0.403
Total Output Tokens
63,667
Total Input Tokens
87,601
Input Price
$1.400 / 1M
Output Price
$4.400 / 1M
Flaky tests
7
Flaky tests had mixed outcomes across runs (at least one pass and one fail).
Hamster playing table tennis
Prompt: Create a detailed SVG illustration of a hamster playing table tennis.
Charts
Choose the first model, then click a second model to open a side-by-side page.
Top Models by Score
Score vs Total Cost
Response Time (avg)
Score vs Response Time (avg)
Total Output Tokens
Score vs Total Output Tokens
Quick Compare
GLM 5.3highvsClaude Sonnet 4.6noneGLM 5.3highvsKAT-Coder-Pro V2.5lowGLM 5.3highvsGemini 3 Flash PreviewlowGLM 5.3highvsGemini 3.1 Flash LitemediumGLM 5.3highvsGrok 4.6lowGLM 5.3highvsQwen3.7 PlusnoneGLM 5.3highvsQwen3.8 2.4T A95BhighGLM 5.3highvsDeepSeek V4 Flash 0731mediumGLM 5.3highvsLongCat 2.0mediumGLM 5.3highvsClaude Opus 4.8none
Category Breakdown
| Category | Score | Consistency | Tests Correct |
|---|---|---|---|
| Anti-AI Tricks | 10.0 | 10.0 | |
| Coding | 6.6 | 5.0 | |
| Combined | 6.4 | 5.8 | |
| Data parsing and extraction | 10.0 | 10.0 | |
| Domain specific | 2.9 | 4.4 | |
| General Intelligence | 6.1 | 3.1 | |
| Instructions following | 9.8 | 10.0 | |
| Puzzle Solving | 8.2 | 7.2 | |
| Tool Calling | 10.0 | 10.0 | |
| Trivia | 3.0 | 10.0 |