Model comparison
GLM-5.2 vs Grok 4
GLM-5.2 is the stronger model overall, scoring 51.1 to 48.1 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 7 categories and Grok 4 in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in long context, where Grok 4 leads 63.1 to 45.3.
- The biggest single-benchmark swing is WeirdML: 70.1% for GLM-5.2 and 45.7% for Grok 4.
- GLM-5.2 has downloadable open weights; the other is API-only.
Side by side
| GLM-5.2 | Grok 4 | |
|---|---|---|
| Provider | Z.ai (Zhipu) | xAI |
| Noometry Index | 51.1 | 48.1 |
| Released | 2026-06-13 | 2025-07-09 |
| Weights | Open | Proprietary |
| Context window | 1M | — |
| Max output | 131K | — |
| Input $ / M tokens | $1.40 | — |
| Output $ / M tokens | $4.40 | — |
| Results tracked | 51 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
GLM-5.2: 51.3 (#41), Grok 4: 50.3 (#46)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| WeirdML | 70.1% | 45.7% |
| LMArena Coding | 1485 | 1408 |
| SWE-bench Verified | 78.7% | — |
| DeepSWE | 43.8% | — |
| FrontierCode | 24.5% | — |
| Aider Polyglot | — | 79.6% |
| LMArena WebDev | 1603 | — |
| SciCode | 50.5% | — |
| ALE-Bench | 1,047 | — |
Agentic & Tool Use Too close to call
GLM-5.2: 32.4 (#63), Grok 4: 32.3 (#68)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| Terminal-Bench | — | 27.2% |
| APEX-Agents | 45.2% | — |
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 21.1% |
| τ²-bench Banking | 37.1% | — |
| Cybench | — | 43% |
| DeepResearch Bench | — | 47.3% |
| PostTrainBench | 31.7% | — |
| BALROG | — | 43.6% |
| GBAEval | 0% | — |
| LMArena Search | — | 1142 |
| METR Time Horizons | — | 66.6% |
| Vending-Bench 2 | 8,314 | — |
Reasoning GLM-5.2 leads
GLM-5.2: 42.3 (#52), Grok 4: 36.7 (#65)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| ARC-AGI-2 | 22.8% | 16% |
| SimpleBench | 58.8% | 60.5% |
| Kagi LLM Benchmark | 62.6% | 73.6% |
| ARC-AGI-1 | 77% | 66.7% |
| Chess Puzzles | 21% | 28% |
| LMArena Hard Prompts | 1480 | 1409 |
| Epoch Capabilities Index | 151.78 | 146.44 |
| NYT Connections (extended) | 74.3% | — |
| CritPt | 20.9% | — |
| EBR-Bench | 9.5% | — |
| Mystery Game Puzzles | 19% | — |
| DTBench | 93.6% | — |
| LMCA | 45.8% | — |
| Surface Evolver Bench | 55.6% | — |
| ForecastBench | — | 60.9 |
Math GLM-5.2 leads
GLM-5.2: 55.7 (#43), Grok 4: 48.4 (#64)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 86.4% | 84% |
| LMArena Math | 1482 | 1422 |
| FrontierMath (Tiers 1-3) | 59.2% | — |
| FrontierMath Tier 4 | 29.3% | — |
| MathArena Final-Answer Competitions | 67.6% | — |
| ProofBench | 35% | — |
| Omni-MATH | — | 60.3% |
| FrontierMath (Feb 2025 set) | — | 19.7% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge GLM-5.2 leads
GLM-5.2: 57.1 (#40), Grok 4: 53.8 (#55)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| GPQA Diamond | 91.9% | 87% |
| LMArena Expert | 1486 | 1415 |
| SimpleQA Verified | 34.2% | — |
| MMLU-Pro | — | 85.1% |
| Confabulations | — | 12.4% |
| GPQA (HELM) | — | 72.7% |
Multimodal Not comparable
GLM-5.2: —, Grok 4: 33.7 (#94)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| LMArena Vision | — | 1210 |
| GeoBench | — | 45% |
Multilingual GLM-5.2 leads
GLM-5.2: 55.8 (#26), Grok 4: 51.8 (#103)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| LMArena Non-English | 1459 | 1403 |
| LMArena Chinese | 1519 | 1427 |
| LMArena French | 1479 | 1418 |
| LMArena German | 1468 | 1429 |
| LMArena Japanese | 1451 | 1394 |
| LMArena Korean | 1445 | 1377 |
| LMArena Russian | 1466 | 1410 |
| LMArena Spanish | 1477 | 1420 |
Instruction Following Grok 4 leads
GLM-5.2: 76.9 (#34), Grok 4: 79.2 (#5)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| LMArena Instruction Following | 1465 | 1387 |
| IFEval | — | 94.9% |
Long Context Grok 4 leads
GLM-5.2: 45.3 (#43), Grok 4: 63.1 (#4)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| LMArena Longer Query | 1479 | 1409 |
| Fiction.LiveBench | — | 94.4% |
Writing & Preference GLM-5.2 leads
GLM-5.2: 70.4 (#21), Grok 4: 58.5 (#116)
| Benchmark | GLM-5.2 | Grok 4 |
|---|---|---|
| LMArena Text | 1470 | 1411 |
| LMArena Creative Writing | 1462 | 1397 |
| LMArena Multi-Turn | 1469 | 1416 |
| Short-Story Creative Writing | — | 76.9% |
| EQ-Bench Creative Writing | 1757 | — |
| WildBench | — | 79.7% |
| EQ-Bench 4 | 1222 | — |
Frequently asked questions
Is GLM-5.2 better than Grok 4?
GLM-5.2 is the stronger model overall, scoring 51.1 to 48.1 on the Noometry Index.
Is GLM-5.2 or Grok 4 better for coding?
They score almost the same on coding (51.3 vs 50.3); test both on your own repository before choosing.
How many benchmarks do GLM-5.2 and Grok 4 share?
26 benchmarks have published results for both models. GLM-5.2 has 51 scored results on Noometry and Grok 4 has 48.