Side-by-side comparison · models
GLM 4.5 Air (Thinking) vs GLM-4.7
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 (Thinking)
Efficient GLM model for fast reasoning, coding, and agent workflows.
Models
GLM-4.7
GLM vision model for visual reasoning, documents, and multimodal agents.
Checked facts
Line by line
| Measure | GLM 4.5 Air (Thinking) | GLM-4.7 |
|---|---|---|
| Input price | Not checked yet | $0.600 per million tokens |
| Output price | Not checked yet | $2.20 per million tokens |
| Price on deepinfra | Not checked yet | $0.064 per million cached input tokens |
| Price on deepinfra | Not checked yet | $0.320 per million input tokens |
| Price on deepinfra | Not checked yet | $1.40 per million output tokens |
| Price on google-vertex | Not checked yet | $0.600 per million input tokens |
| Price on google-vertex | Not checked yet | $2.20 per million output tokens |
| Price on mancer | Not checked yet | $0.350 per million cached input tokens |
| Price on mancer | Not checked yet | $0.700 per million input tokens |
| Price on mancer | Not checked yet | $2.50 per million output tokens |
| Price on novita | Not checked yet | $0.110 per million cached input tokens |
| Price on novita | Not checked yet | $0.600 per million input tokens |
| Price on novita | Not checked yet | $2.20 per million output tokens |
| Price on venice | Not checked yet | $0.080 per million cached input tokens |
| Price on venice | Not checked yet | $0.400 per million input tokens |
| Price on venice | Not checked yet | $1.93 per million output tokens |
| Price on z-ai | Not checked yet | $0.110 per million cached input tokens |
| Price on z-ai | Not checked yet | $0.600 per million input tokens |
| Price on z-ai | Not checked yet | $2.20 per million output tokens |
| Latest version | Not checked yet | z-ai/glm-4.7-20251222 |
| Time to first token (ms) | 2,390 | 2,510 |
| Output speed (tokens a second) | 82.8 | 39.5 |
| Time to first token (ms) on deepinfra | Not checked yet | 2,091 |
| Time to first token (ms) on mancer | Not checked yet | 1,060 |
| Time to first token (ms) on novita | Not checked yet | 4,540 |
| Time to first token (ms) on venice | Not checked yet | 1,395 |
| Time to first token (ms) on z-ai | Not checked yet | 3,922 |
| Output speed (tokens a second) on deepinfra | Not checked yet | 36 |
| Output speed (tokens a second) on mancer | Not checked yet | 27.5 |
| Output speed (tokens a second) on novita | Not checked yet | 24 |
| Output speed (tokens a second) on venice | Not checked yet | 30 |
| Output speed (tokens a second) on z-ai | Not checked yet | 27 |
| Tokens served on OpenRouter in a day | Not checked yet | 4.6B (for 10 Nov 2026) |
| Tool calls on OpenRouter in a day | Not checked yet | 23.8K (for 10 Nov 2026) |
| Artificial Analysis Intelligence Index | Not checked yet | 22.2 |
| Artificial Analysis Coding Index | Not checked yet | 45.3 |
| LMArena text rating (Elo) | Not checked yet | 1,435 |
| LMArena web development rating (Elo) | Not checked yet | 1,435 |
| Design Arena rating (Elo) | Not checked yet | 1,222 |
| Gpqa diamond | Not checked yet | 83.3% (for 22 Dec 2025) |
| Terminal bench | Not checked yet | 33.4% (for 22 Dec 2025) |
| Simpleqa verified | Not checked yet | 32.2% (for 22 Dec 2025) |
| Chess puzzles | Not checked yet | 21% (for 22 Dec 2025) |
| Cl bench | Not checked yet | 15.9% (for 22 Dec 2025) |
| Cl bench life | Not checked yet | 10.9% (for 22 Dec 2025) |
| Frontiermath 2025 02 28 private | Not checked yet | 2.4% (for 22 Dec 2025) |
| Frontiermath tier 4 2025 07 01 private | Not checked yet | 0% (for 22 Dec 2025) |
| Proofbench | Not checked yet | 6% (for 22 Dec 2025) |
| Simplebench | Not checked yet | 47.7% (for 22 Dec 2025) |
| Otis mock aime 2024 2025 | Not checked yet | 83.3% (saturated benchmark: the top models all score near the ceiling) (for 22 Dec 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.