Model comparison
GLM-5.2 vs GPT-4
GLM-5.2 is the stronger model overall, scoring 51.1 to 29.1 on the Noometry Index.
Last verified . 26 shared benchmarks.
Summary
- They share 26 benchmarks with published results for both. GLM-5.2 scores higher in 8 categories and GPT-4 in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.2 leads 55.7 to 10.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 86.4% for GLM-5.2 and 1.1% for GPT-4.
- GLM-5.2 is cheaper at $1.40 / $4.40 per million input/output tokens, against $30 / $60 for GPT-4.
- GLM-5.2 accepts more context: 1M tokens versus 8K.
- GLM-5.2 has downloadable open weights; the other is API-only.
Side by side
| GLM-5.2 | GPT-4 | |
|---|---|---|
| Provider | Z.ai (Zhipu) | OpenAI |
| Noometry Index | 51.1 | 29.1 |
| Released | 2026-06-13 | 2023-03-14 |
| Weights | Open | Proprietary |
| Context window | 1M | 8K |
| Max output | 131K | 8K |
| Input $ / M tokens | $1.40 | $30 |
| Output $ / M tokens | $4.40 | $60 |
| Results tracked | 51 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.2 leads
GLM-5.2: 51.3 (#41), GPT-4: 31.6 (#283)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| WeirdML | 70.1% | 12.4% |
| LMArena Coding | 1485 | 1254 |
| SWE-bench Verified | 78.7% | — |
| DeepSWE | 43.8% | — |
| FrontierCode | 24.5% | — |
| LMArena WebDev | 1603 | — |
| SciCode | 50.5% | — |
| BigCodeBench Instruct | — | 46% |
| BigCodeBench Complete | — | 57.2% |
| ALE-Bench | 1,047 | — |
| HumanEval+ | — | 79.3% |
Agentic & Tool Use Not comparable
GLM-5.2: 32.4 (#63), GPT-4: —
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| APEX-Agents | 45.2% | — |
| τ²-bench Banking | 37.1% | — |
| PostTrainBench | 31.7% | — |
| GBAEval | 0% | — |
| METR Time Horizons | — | 36.1% |
| Vending-Bench 2 | 8,314 | — |
Reasoning GLM-5.2 leads
GLM-5.2: 42.3 (#52), GPT-4: 17.8 (#289)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| Chess Puzzles | 21% | 4% |
| LMArena Hard Prompts | 1480 | 1241 |
| Mystery Game Puzzles | 19% | 12% |
| DTBench | 93.6% | 62.7% |
| LMCA | 45.8% | 17.1% |
| Epoch Capabilities Index | 151.78 | 125.89 |
| ARC-AGI-2 | 22.8% | — |
| SimpleBench | 58.8% | — |
| Kagi LLM Benchmark | 62.6% | — |
| NYT Connections (extended) | 74.3% | — |
| ARC-AGI-1 | 77% | — |
| CritPt | 20.9% | — |
| EBR-Bench | 9.5% | — |
| Surface Evolver Bench | 55.6% | — |
| BIG-Bench Hard | — | 75.1% |
| ForecastBench | — | 57.8 |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |
Math GLM-5.2 leads
GLM-5.2: 55.7 (#43), GPT-4: 10.8 (#309)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 86.4% | 1.1% |
| LMArena Math | 1482 | 1269 |
| FrontierMath (Tiers 1-3) | 59.2% | — |
| FrontierMath Tier 4 | 29.3% | — |
| MathArena Final-Answer Competitions | 67.6% | — |
| ProofBench | 35% | — |
| MATH Level 5 | — | 23% |
| GSM8K | — | 92% |
Knowledge GLM-5.2 leads
GLM-5.2: 57.1 (#40), GPT-4: 18.4 (#282)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| GPQA Diamond | 91.9% | 35.7% |
| LMArena Expert | 1486 | 1211 |
| SimpleQA Verified | 34.2% | — |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |
Multilingual GLM-5.2 leads
GLM-5.2: 55.8 (#26), GPT-4: 40.6 (#215)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| LMArena Non-English | 1459 | 1246 |
| LMArena Chinese | 1519 | 1242 |
| LMArena French | 1479 | 1283 |
| LMArena German | 1468 | 1251 |
| LMArena Japanese | 1451 | 1209 |
| LMArena Korean | 1445 | 1184 |
| LMArena Russian | 1466 | 1251 |
| LMArena Spanish | 1477 | 1261 |
Instruction Following GLM-5.2 leads
GLM-5.2: 76.9 (#34), GPT-4: 65.3 (#222)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1465 | 1241 |
Long Context GLM-5.2 leads
GLM-5.2: 45.3 (#43), GPT-4: 37.7 (#212)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| LMArena Longer Query | 1479 | 1244 |
Writing & Preference GLM-5.2 leads
GLM-5.2: 70.4 (#21), GPT-4: 34.9 (#268)
| Benchmark | GLM-5.2 | GPT-4 |
|---|---|---|
| LMArena Text | 1470 | 1263 |
| LMArena Creative Writing | 1462 | 1244 |
| EQ-Bench Creative Writing | 1757 | 752 |
| LMArena Multi-Turn | 1469 | 1257 |
| EQ-Bench 4 | 1222 | — |
Frequently asked questions
Is GLM-5.2 better than GPT-4?
GLM-5.2 is the stronger model overall, scoring 51.1 to 29.1 on the Noometry Index.
Which is cheaper, GLM-5.2 or GPT-4?
GLM-5.2 is cheaper. It lists at $1.40 per million input tokens and $4.40 per million output tokens; GPT-4 lists at $30 and $60.
Is GLM-5.2 or GPT-4 better for coding?
GLM-5.2 scores higher on coding benchmarks: 51.3 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
GLM-5.2 does, with 1M tokens against 8K.
How many benchmarks do GLM-5.2 and GPT-4 share?
26 benchmarks have published results for both models. GLM-5.2 has 51 scored results on Noometry and GPT-4 has 38.