GLM-5 is Z.ai's flagship open-source foundation model, scaling to 744B total parameters (40B active) and trained on 28.5T tokens. It is engineered for complex systems design and long-horizon agent workflows, delivering production-grade performance on large-scale programming tasks for expert developers.
Cost rate
0.7x
Context
205K
Released
Feb 11, 2026
Input
Text
Output
Text
Support
ReasoningTool callingStructured outputs
Performance
Median latency and throughput measured across recent requests.
Throughput58 tok/s
Latency3.2s
Best at
The categories where this model ranks highest.
Entertainment, Sports, & Media
#58 · top 14%
Medicine & Healthcare
#62 · top 16%
Writing, Literature, & Language
#65 · top 15%
Life, Physical, & Social Science
#67 · top 16%
| Domain | Score | Rank |
|---|---|---|
| CodingCode Arena Elo | 1435 | #97 |
| Software & IT Services | 1488 | #84 |
| Writing, Literature, & Language | 1447 | #65 |
| Life, Physical, & Social Science | 1479 | #67 |
| Entertainment, Sports, & Media | 1438 | #58 |
| Business, Management, & Finance | 1451 | #79 |
| Mathematical | 1455 | #84 |
| Legal & Government | 1468 | #67 |
| Medicine & Healthcare | 1479 | #62 |
| Category | Win rate | Rank |
|---|---|---|
| Code Categories | 55.5% | #48 |
| Game Development | 57.4% | #48 |
| SVG | 54.4% | #48 |
| Data Visualization | 53.0% | #49 |
| ASCII Art | 47.9% | #50 |
| 3D | 56.3% | #51 |
| Website | 55.0% | #53 |
| UI Component | 53.6% | #59 |
Latest news about GLM 5
Recent articles and community posts, linked to their original sources.
- Fixstars adds GLM-5.3 to its Japan-hosted AI inference serviceInternational Business Times



GLM 5.5 32B leaked. Don’t wait for it. #AIRun Local AI2,160
- GLM-LegalHacker News676371
How GLM built its own inference infrastructureHacker News411285