Side-by-side comparison · models
GLM 4.5 Air vs GLM 4.6
What we have checked about each, line by line. Every value shows where it came from and when we last confirmed it.
Models
GLM 4.5 Air
GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications.
Models
GLM 4.6
Compared with GLM-4.5, this generation brings several key improvements: Longer context window: The context window has been expanded from 128K to 200K tokens, enabling the model.
Checked facts
Line by line
| Measure | GLM 4.5 Air | GLM 4.6 |
|---|---|---|
| Input price | $0.200 per million tokens | $0.500 per million tokens |
| Output price | $1.10 per million tokens | $2.00 per million tokens |
| Price on novita | $0.025 per million cached input tokens | $0.110 per million cached input tokens |
| Price on novita | $0.130 per million input tokens | $0.550 per million input tokens |
| Price on novita | $0.850 per million output tokens | $2.20 per million output tokens |
| Price on siliconflow | $0.140 per million input tokens | Not checked yet |
| Price on siliconflow | $0.860 per million output tokens | Not checked yet |
| Price on z-ai | $0.030 per million cached input tokens | $0.110 per million cached input tokens |
| Price on z-ai | $0.200 per million input tokens | $0.600 per million input tokens |
| Price on z-ai | $1.10 per million output tokens | $2.20 per million output tokens |
| Price on deepinfra | Not checked yet | $0.080 per million cached input tokens |
| Price on deepinfra | Not checked yet | $0.400 per million input tokens |
| Price on deepinfra | Not checked yet | $1.60 per million output tokens |
| Price on venice | Not checked yet | $0.080 per million cached input tokens |
| Price on venice | Not checked yet | $0.430 per million input tokens |
| Price on venice | Not checked yet | $1.75 per million output tokens |
| Latest version | z-ai/glm-4.5-air | z-ai/glm-4.6 |
| Time to first token (ms) | 2,390 | 2,860 |
| Time to first token (ms) on novita | 720 | 5,070 |
| Time to first token (ms) on siliconflow | 1,085 | Not checked yet |
| Time to first token (ms) on z-ai | 982 | 4,378 |
| Output speed (tokens a second) | 82.8 | 36.4 |
| Output speed (tokens a second) on novita | 54 | 27 |
| Output speed (tokens a second) on siliconflow | 53 | Not checked yet |
| Output speed (tokens a second) on z-ai | 31 | 28 |
| Time to first token (ms) on deepinfra | Not checked yet | 3,826 |
| Time to first token (ms) on venice | Not checked yet | 1,449 |
| Output speed (tokens a second) on deepinfra | Not checked yet | 34 |
| Output speed (tokens a second) on venice | Not checked yet | 30 |
| Tokens served on OpenRouter in a day | 430.8M (for 10 Nov 2026) | 2.3B (for 10 Nov 2026) |
| Tool calls on OpenRouter in a day | 5.8K (for 10 Nov 2026) | 35.9K (for 10 Nov 2026) |
| Artificial Analysis Intelligence Index | 11.1 | 18.5 |
| LMArena text rating (Elo) | 1,384 | 1,440 |
| Design Arena rating (Elo) | 1,184 | 1,180 |
| Gso bench | 2.9% (for 20 July 2025) | Not checked yet |
| Artificial Analysis Coding Index | Not checked yet | 45.8 |
| LMArena web development rating (Elo) | Not checked yet | 1,342 |
| Terminal bench | Not checked yet | 24.5% (for 30 Sept 2025) |
| Frontiermath 2025 02 28 private | Not checked yet | 3.8% (for 30 Sept 2025) |
| Frontiermath tier 4 2025 07 01 private | Not checked yet | 2.1% (for 30 Sept 2025) |
A line appears only when at least one record has a value we confirmed from a source. Values we have not confirmed are left out, never guessed. Records of different types share fewer lines.
Add a third
Compare with one more record
How this works
Share it, or build your own
This page works for any two or three records in the directory, of any type. The link stays the same whichever order you name them in, so you can share it as it is. Change the records in the address, for example /directories/vs/claude-code-vs-cursor, or start from our editorial comparisons.