Loading page content
GLM-5.3Closed
Tool use · Reasoning
Evaluation overview
Overall rank
#3
98.9 points · 194 models
Coding rank
#2
94.6 points · 174 models
Data coverage
17 items
1 source · updated through 2026-08-14
Domain performance
Coding#2 / 174
94.6
Research#4 / 188
99
Agents#3 / 140
93.8
Evaluation highlights
All evaluationsBasic information
- Context (input / output)
- 1.05M / 131K
- Parameters
- 744B / 40B active
- Release date
- 2026-08
- Input modalities
- Text
- Output modalities
- Text
- Exact release date
- 2026-08-14
- Base model
- Same as GLM-5.2 (all gains come from post-training)
- Architecture
- MoE (256 experts), DSA, IndexShare
- Availability
- GLM Coding Plan, ZCode, and API
- Reasoning mode
- Thinking only (supports low / high / max)
- Weight status
- Open weights planned for August 28, 2026, after safety evaluation and hardening
- Evaluation source
- Official 16-row comparison table, split into 17 metrics and 120 scores
- API model ID
- glm-5.3
- Default reasoning effort
- max (supports low / high / max)
Related links
Version relationships
GLM-5 family
GLM-5
2026-02 · Same series
GLM-5.1
2026-04 · Same series
GLM-5.2
2026-06 · Predecessor
GLM-5.3Current
2026-08 · Current model