Model comparison
GLM-5.2 vs Grok 4.20 Multi-Agent
GLM-5.2 is the stronger model overall, scoring 50.9 to 45.5 on the Noometry Index.
Last verified . 18 shared benchmarks.
Summary
- They share 18 benchmarks with published results for both. GLM-5.2 scores higher in 7 categories and Grok 4.20 Multi-Agent in 1 category; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.2 leads 55.4 to 36.6.
- The biggest single-benchmark swing is NYT Connections (extended): 74.3% for GLM-5.2 and 89.6% for Grok 4.20 Multi-Agent.
- Grok 4.20 Multi-Agent is cheaper at $1.25 / $2.50 per million input/output tokens, against $1.40 / $4.40 for GLM-5.2.
- GLM-5.2 has downloadable open weights; the other is API-only.
Side by side
| GLM-5.2 | Grok 4.20 Multi-Agent | |
|---|---|---|
| Provider | Z.ai (Zhipu) | xAI |
| Noometry Index | 50.9 | 45.5 |
| Released | 2026-06-13 | 2026-03-09 |
| Weights | Open | Proprietary |
| Context window | 1M | 1M |
| Max output | 131K | 30K |
| Input $ / M tokens | $1.40 | $1.25 |
| Output $ / M tokens | $4.40 | $2.50 |
| Results tracked | 50 | 20 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.2 leads
GLM-5.2: 50.9 (#37), Grok 4.20 Multi-Agent: 42.4 (#91)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Coding | 1484 | 1458 |
| SWE-bench Verified | 78.7% | — |
| DeepSWE | 43.8% | — |
| FrontierCode | 24.5% | — |
| LMArena WebDev | 1604 | — |
| SciCode | 50.5% | — |
| WeirdML | 70.1% | — |
| ALE-Bench | 1,047 | — |
Agentic & Tool Use Not comparable
GLM-5.2: 31.6 (#63), Grok 4.20 Multi-Agent: —
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| τ²-bench Banking | 37.1% | — |
| PostTrainBench | 31.7% | — |
| GBAEval | 0% | — |
| LMArena Search | — | 1204 |
| Vending-Bench 2 | 8,314 | — |
Reasoning Too close to call
GLM-5.2: 42.1 (#46), Grok 4.20 Multi-Agent: 42.4 (#44)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| NYT Connections (extended) | 74.3% | 89.6% |
| LMArena Hard Prompts | 1480 | 1448 |
| ARC-AGI-2 | 22.8% | — |
| SimpleBench | 58.8% | — |
| Kagi LLM Benchmark | 62.6% | — |
| ARC-AGI-1 | 77% | — |
| CritPt | 20.9% | — |
| Chess Puzzles | 21% | — |
| EBR-Bench | 9.5% | — |
| Mystery Game Puzzles | 19% | — |
| DTBench | 93.6% | — |
| LMCA | 45.8% | — |
| Surface Evolver Bench | 55.6% | — |
| Epoch Capabilities Index | 151.76 | — |
Math GLM-5.2 leads
GLM-5.2: 55.4 (#35), Grok 4.20 Multi-Agent: 36.6 (#106)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Math | 1482 | 1441 |
| FrontierMath (Tiers 1-3) | 59.2% | — |
| FrontierMath Tier 4 | 29.3% | — |
| MathArena Final-Answer Competitions | 67.6% | — |
| OTIS Mock AIME 2024-2025 | 86.4% | — |
| ProofBench | 35% | — |
Knowledge GLM-5.2 leads
GLM-5.2: 57.1 (#36), Grok 4.20 Multi-Agent: 39.7 (#116)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Expert | 1484 | 1445 |
| GPQA Diamond | 91.9% | — |
| SimpleQA Verified | 34.2% | — |
Multimodal Not comparable
GLM-5.2: —, Grok 4.20 Multi-Agent: 39.8 (#44)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Vision | — | 1259 |
Multilingual GLM-5.2 leads
GLM-5.2: 56.0 (#25), Grok 4.20 Multi-Agent: 54.6 (#40)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Non-English | 1459 | 1440 |
| LMArena Chinese | 1518 | 1475 |
| LMArena French | 1476 | 1467 |
| LMArena German | 1470 | 1456 |
| LMArena Japanese | 1451 | 1405 |
| LMArena Korean | 1445 | 1416 |
| LMArena Russian | 1467 | 1457 |
| LMArena Spanish | 1477 | 1447 |
Instruction Following GLM-5.2 leads
GLM-5.2: 77.1 (#32), Grok 4.20 Multi-Agent: 75.0 (#80)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Instruction Following | 1465 | 1420 |
Long Context GLM-5.2 leads
GLM-5.2: 45.3 (#43), Grok 4.20 Multi-Agent: 43.7 (#84)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Longer Query | 1479 | 1430 |
Writing & Preference GLM-5.2 leads
GLM-5.2: 70.7 (#20), Grok 4.20 Multi-Agent: 64.1 (#57)
| Benchmark | GLM-5.2 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Text | 1470 | 1450 |
| LMArena Creative Writing | 1465 | 1435 |
| LMArena Multi-Turn | 1468 | 1451 |
| EQ-Bench Creative Writing | 1757 | — |
| EQ-Bench 4 | 1222 | — |
Frequently asked questions
Is GLM-5.2 better than Grok 4.20 Multi-Agent?
GLM-5.2 is the stronger model overall, scoring 50.9 to 45.5 on the Noometry Index.
Which is cheaper, GLM-5.2 or Grok 4.20 Multi-Agent?
Grok 4.20 Multi-Agent is cheaper. It lists at $1.25 per million input tokens and $2.50 per million output tokens; GLM-5.2 lists at $1.40 and $4.40.
Is GLM-5.2 or Grok 4.20 Multi-Agent better for coding?
GLM-5.2 scores higher on coding benchmarks: 50.9 versus 42.4 in the Noometry coding category.
Which has the bigger context window?
Both accept 1M tokens.
How many benchmarks do GLM-5.2 and Grok 4.20 Multi-Agent share?
18 benchmarks have published results for both models. GLM-5.2 has 50 scored results on Noometry and Grok 4.20 Multi-Agent has 20.