GLM-5.2 vs GLM-5.3
GLM-5.2 (Z.ai (Zhipu AI)) and GLM-5.3 (Z.ai (Zhipu AI)) compared on benchmarks, pricing, context window, and use-case rankings.
- GLM-5.3 leads on overall intelligence (ECI 157 vs 153).
- GLM-5.3 is cheaper on input tokens: $0.40 vs $0.70 per 1M.
- GLM-5.3 ranks higher for coding (89th vs 74th percentile).
| Attribute | GLM-5.2 | GLM-5.3 |
|---|---|---|
| Scores | ||
| Intelligence (ECI) | 153 | 157 |
| Coding | 74 | 89 |
| Math | 20 | 35 |
| Reasoning & Knowledge | 43 | 62 |
| Agentic & Tools | 73 | 73 |
| Specifications | ||
| Developer | Z.ai (Zhipu AI) | Z.ai (Zhipu AI) |
| Family | — | — |
| Released | Jun 16, 2026 | Aug 14, 2026 |
| Parameters | — | 744B |
| Availability | Open weights (unrestricted) | Open weights (restricted use) |
| Context window | 1M | 1M |
| Price — $/M input | $0.70 | $0.40 |
| Price — $/M output | $2.20 | $2.46 |
| Inputs | text | text |
| Outputs | text | text |
| Benchmarks | ||
| AIME 2024/2025 | 86% | 91% |
| APEX | 45% | 57% |
| ARC-AGI | 77% | — |
| ARC-AGI-2 | 23% | — |
| GPQA Diamond | 92% | 91% |
| Humanity's Last Exam | 41% | 42% |
| SciCode | 51% | 59% |
| SimpleBench | 59% | 66% |
| SimpleQA Verified | 34% | 41% |
| SWE-bench Verified | 79% | — |
| WebDev Arena | 1593 | 1614 |
| WeirdML | 70% | 75% |
| τ²-bench | 99% | — |
Use-case scores are 0–100 percentile composites across each area’s benchmarks, ranked against every model from the past year. Highlighted cells lead each row. Open a model for the full picture.
Frequently asked questions
Is GLM-5.2 better than GLM-5.3?
On Epoch AI's Capabilities Index, GLM-5.3 scores higher (157) than GLM-5.2 (153). The right pick depends on your task, so compare their coding, math, and reasoning scores in the table above.
Which is cheaper, GLM-5.2 or GLM-5.3?
GLM-5.3 is cheaper on input tokens at $0.40 per million, versus $0.70 (representative OpenRouter pricing).
Which has a larger context window, GLM-5.2 or GLM-5.3?
Neither: both support up to 1M tokens.
Which is better for coding, GLM-5.2 or GLM-5.3?
GLM-5.3 ranks higher across coding benchmarks like WebDev Arena and SciCode: 89th vs 74th percentile among the models tracked on Model Beat.
Want a different match-up? Open the compare tool to add or swap models.
More comparisons
Benchmarks & model data from Epoch AI (CC BY); pricing & specs from OpenRouter. ECI = Epoch Capabilities Index.