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: 编程与软件工程 +10.1 / Relative gap: none clear
Relative edge: none clear / Relative gap: 编程与软件工程 -10.1
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
GLM-5.2 · 68.71
Best single
GLM-5.2 · GPQA Diamond 91.86
Modality coverage
GLM-5.2 · 1 modalities
Head to head
8
Benchmarks
6
Wins
2
Losses
+3.86
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.
8 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.
| Benchmark | GLM-5.2 | Hy3 |
|---|---|---|
HLE 综合评估 | 54.70Thinking Enabled ｜ Tools | 53.20Thinking Level · High ｜ Tools |
GPQA Diamond 科学与综合推理 | 91.86Thinking Level · High | 90.40Thinking Level · High |
DeepSWE 编程与软件工程 | 44.00Deep Thinking Mode ｜ Tools | 28.00Thinking Level · High ｜ Tools |
SWE-Bench Pro - Public 编程与软件工程 | 62.10Thinking Enabled ｜ Tools | 57.90Thinking Level · High ｜ Tools |
MCP-Atlas AI Agent - 工具使用 | 76.80Thinking Enabled ｜ Tools | 79.10Thinking Level · High ｜ Tools |
Terminal-Bench 2.1 AI Agent - 工具使用 | 81.00Thinking Level · High ｜ Tools | 71.70Thinking Level · High ｜ Tools |
Tool Decathlon AI Agent - 工具使用 | 48.20Thinking Enabled ｜ Tools | 48.50Thinking Level · High ｜ Tools |
IMO-AnswerBench 数学推理 | 91.00Thinking Enabled | 90.00Thinking Level · High |
Side-by-side input/output token pricing
Licensing, MoE architecture, and multi-modality support.
| Features & specs | GLM-5.2智谱AI | Hy3腾讯AI实验室 |
|---|---|---|
Core specsRelease | 2026-06-13 | 2026-07-06 |
Context length | 1M | 256K |
Parameters | 753.3B | 295B |
Active parameters | 40B | 21B |
Max output | 128000 | Not provided |
MoE | Yes | Yes |
LicenseCode Open Source | Open Source · MIT License | Open Source · Apache 2.0 |
Weights Open Source | Open Source · MIT License | Open Source · Apache 2.0 |
Commercial use | 免费商用授权 | 免费商用授权 |
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 | Hugging Face |
Source repo | GitHub | GitHub |
Modality supportText Input/Output | / | / |
ResourcesPaper / report | GLM-5: from Vibe Coding to Agentic Engineering | Hy3 |
