See key specs and per-benchmark scores for each model/mode. Scroll horizontally for all columns. 当前对比 2 个模型的评测数据与核心参数。
GLM-5.2
智谱AI
Each axis is a category average, normalized to a 100-point radar.
Relative edge: 数学推理 +3.5 / Relative gap: 编程与软件工程 -9.2
Relative edge: 编程与软件工程 +9.2 / Relative gap: 数学推理 -3.5
Method: for each model and benchmark, the chart first averages all scores in the current mode scope instead of taking the best score, then averages those benchmark scores within each category. Only benchmarks with at least two selected models scored are included; missing values are not counted as zero.
Best overall
Grok 4.5 · 57.86
Best single
Grok 4.5 · GPQA Diamond 93.43
Modality coverage
Grok 4.5 · 2 modalities
Head to head
7
Benchmarks
2
Wins
5
Losses
-3.51
Average diff
Compare benchmark results across thinking modes and tool usage.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Complete scores for each model/mode across selected benchmarks.
7 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.
| Benchmark | GLM-5.2 | Grok 4.5 |
|---|---|---|
GPQA Diamond 科学与综合推理 | 91.86Thinking Level · High | 93.43Thinking Level · High |
DeepSWE 编程与软件工程 | 44.00Deep Thinking Mode ｜ Tools | 53.00Thinking Level · High ｜ Tools |
SWE-Bench Pro - Public 编程与软件工程 | 62.10Thinking Enabled ｜ Tools | 64.70Thinking Level · High ｜ Tools |
SWE-Marathon 编程与软件工程 | 13.00Thinking Level · High ｜ Tools | 29.00Thinking Level · High ｜ Tools |
Terminal-Bench 2.1 AI Agent - 工具使用 | 81.00Thinking Level · High ｜ Tools | 83.30Thinking Level · High ｜ Tools |
29.27Thinking Level · High | 24.39Thinking Level · High | |
FrontierMath v2 数学推理 | 59.21Thinking Level · High | 57.19Thinking Level · High |
Side-by-side input/output token pricing
Licensing, MoE architecture, and multi-modality support.
| Features & specs | GLM-5.2智谱AI | Grok 4.5xAI |
|---|---|---|
Core specsRelease | 2026-06-13 | 2026-07-08 |
Context length | 1M | 500K |
Parameters | 753.3B | — |
Active parameters | 40B | Not provided |
Max output | 128000 | Not provided |
MoE | Yes | No |
LicenseCode Open Source | Open Source · MIT License | Closed Source |
Weights Open Source | Open Source · MIT License | Closed Source |
Commercial use | 免费商用授权 | 不开源 |
Local deploymentWeight size | 约 1.5 TB（BF16 safetensors） | Not provided |
VRAM for weights | ≈ 1500 GB (weights only, excludes KV cache) | Not provided |
Weights | Hugging Face | Not provided |
Source repo | GitHub | Not provided |
Modality supportText Input/Output | / | / |
Image Input/Output | Not provided | / |
ResourcesPaper / report | GLM-5: from Vibe Coding to Agentic Engineering | Introducing Grok 4.5 |
