GLM-5.2vsDeepSeek-V4-Pro
Across 4 shared benchmarks, GLM-5.2 leads overall: GLM-5.2 wins 4, DeepSeek-V4-Pro wins 0, with 0 ties and an average score difference of +32.75.
GLM-5.2
智谱AI · 2026-06-13 · Reasoning model
DeepSeek-V4-Pro
DeepSeek-AI · 2026-04-24 · Reasoning model
GLM-5.24 wins(100%)(0%)0 winsDeepSeek-V4-Pro
Benchmark scores
Grouped by capability, sorted by largest gap within each. 4 shared benchmarks.
General Knowledge
GLM-5.2 2/2| Benchmark | GLM-5.2 | DeepSeek-V4-Pro | Diff |
|---|---|---|---|
| HLE | 54.708 / 159Thinking (With Tools) | 7.70143 / 159Normal (No Tools) | +47 |
| GPQA Diamond | 91.2015 / 179Thinking (No Tools) | 72.90103 / 179Normal (No Tools) | +18.30 |
Coding and Software Engineer
GLM-5.2 1/1| Benchmark | GLM-5.2 | DeepSeek-V4-Pro | Diff |
|---|---|---|---|
| SWE-Bench Pro - Public | 62.105 / 44Thinking (With Tools) | 52.1029 / 44Normal (With Tools) | +10 |
Math and Reasoning
GLM-5.2 1/1| Benchmark | GLM-5.2 | DeepSeek-V4-Pro | Diff |
|---|---|---|---|
| IMO-AnswerBench | 911 / 20Thinking (No Tools) | 35.3020 / 20Normal (No Tools) | +55.70 |
Specs
| Field | GLM-5.2 | DeepSeek-V4-Pro |
|---|---|---|
| Publisher | 智谱AI | DeepSeek-AI |
| Release date | 2026-06-13 | 2026-04-24 |
| Model type | Reasoning model | Reasoning model |
| Architecture | MoE | MoE |
| Parameters | 753.33B | 1.6T |
| Context length | 1M | 1M |
| Max output | 128K | 375K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GLM-5.2 | DeepSeek-V4-Pro |
|---|---|---|
| Text input | $1.4 / 1M tokens | $0.435 / 1M tokens |
| Text output | $4.4 / 1M tokens | $0.87 / 1M tokens |
| Cache read | $0.26 / 1M tokens | $0.87 / 1M tokens |
| Cache write | Not public | $0.003625 / 1M tokens |
Summary
- GLM-5.2leads in:General Knowledge (2/2), Coding and Software Engineer (1/1), Math and Reasoning (1/1)
On average across the 4 shared benchmarks, GLM-5.2 scores 32.75 higher.
Largest single-benchmark gap: IMO-AnswerBench — GLM-5.2 91 vs DeepSeek-V4-Pro 35.30 (+55.70).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.