Claude Sonnet 5vsGLM-5.2
Claude Sonnet 5 and GLM-5.2 are tied across 9 shared benchmarks: Claude Sonnet 5 leads on 4, GLM-5.2 leads on 4, with 1 ties and an average score difference of -3.48.
Claude Sonnet 5
Anthropic · 2026-06-30 · Multimodal model
GLM-5.2
智谱AI · 2026-06-13 · Reasoning model
Claude Sonnet 54 wins(44%)Ties1(44%)4 winsGLM-5.2
Benchmark scores
Grouped by capability, sorted by largest gap within each. 9 shared benchmarks.
Coding and Software Engineer
Claude Sonnet 5 2/3| Benchmark | Claude Sonnet 5 | GLM-5.2 | Diff |
|---|---|---|---|
| Text Arena (Coding) | 1,54410 / 35Thinking High (No Tools) | 1,5935 / 35最高（无工具） | -49.09 |
| DeepSWE | 5416 / 27Deep Thinking (With Tools) | 4421 / 27Deep Thinking (With Tools) | +10 |
| WeirdML v2 | 68.7820 / 52Thinking High (With Tools) | 67.3121 / 52Thinking High (With Tools) | +1.47 |
Math and Reasoning
Claude Sonnet 5 1/2| Benchmark | Claude Sonnet 5 | GLM-5.2 | Diff |
|---|---|---|---|
| FrontierMath v2 | 65.6117 / 34最高（无工具） | 59.2119 / 34最高（无工具） | +6.41 |
| FrontierMath Tier 4 v2 | 29.2716 / 34最高（无工具） | 29.2716 / 34最高（无工具） | — |
AI Agent - Tool Usage
GLM-5.2 1/1| Benchmark | Claude Sonnet 5 | GLM-5.2 | Diff |
|---|---|---|---|
| Terminal-Bench 2.1 | 80.4018 / 44极高强度思考（工具） | 8117 / 44Thinking High (With Tools) | -0.60 |
Commonsense Reasoning
GLM-5.2 1/1| Benchmark | Claude Sonnet 5 | GLM-5.2 | Diff |
|---|---|---|---|
| SimpleBench | 57.9023 / 67Normal (No Tools) | 58.8020 / 67Normal (No Tools) | -0.90 |
General Evaluation
GLM-5.2 1/1| Benchmark | Claude Sonnet 5 | GLM-5.2 | Diff |
|---|---|---|---|
| GPQA Diamond | 90.5333 / 226极高强度思考（无工具） | 91.8626 / 226最高（无工具） | -1.33 |
General Knowledge
Claude Sonnet 5 1/1| Benchmark | Claude Sonnet 5 | GLM-5.2 | Diff |
|---|---|---|---|
| HLE | 57.409 / 181极高强度思考（工具） | 54.7015 / 181Thinking (With Tools) | +2.70 |
Specs
| Field | Claude Sonnet 5 | GLM-5.2 |
|---|---|---|
| Publisher | Anthropic | 智谱AI |
| Release date | 2026-06-30 | 2026-06-13 |
| Model type | Multimodal model | Reasoning model |
| Architecture | Dense | MoE |
| Parameters | Not available | 753.33B |
| Context length | 1M | 1M |
| Max output | 128K | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | Claude Sonnet 5 | GLM-5.2 |
|---|---|---|
| Text input | $2 / 1M tokens | $1.4 / 1M tokens |
| Text output | $10 / 1M tokens | $4.4 / 1M tokens |
| Cache read | $0.3 / 1M tokens | $0.26 / 1M tokens |
| Cache write | $3.75 / 1M tokens | Not public |
Summary
- Claude Sonnet 5leads in:Coding and Software Engineer (2/3), Math and Reasoning (1/2), General Knowledge (1/1)
- GLM-5.2leads in:AI Agent - Tool Usage (1/1), Commonsense Reasoning (1/1), General Evaluation (1/1)
On average across the 9 shared benchmarks, GLM-5.2 scores 3.48 higher.
Largest single-benchmark gap: Text Arena (Coding) — Claude Sonnet 5 1,544 vs GLM-5.2 1,593 (-49.09).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.