Model comparison
GLM-5.3 vs Qwen2.5 72B Instruct
GLM-5.3 is the stronger model overall, scoring 54.6 to 31.5 on the Noometry Index.
Last verified . 23 shared benchmarks.
Alibaba (Qwen)Summary
- They share 23 benchmarks with published results for both. GLM-5.3 scores higher in 9 categories and Qwen2.5 72B Instruct in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.3 leads 62.1 to 18.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 91.1% for GLM-5.3 and 8.1% for Qwen2.5 72B Instruct.
- GLM-5.3 is cheaper at $1.40 / $4.40 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
- GLM-5.3 accepts more context: 1M tokens versus 131K.
Side by side
| GLM-5.3 | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Z.ai (Zhipu) | Alibaba (Qwen) |
| Noometry Index | 54.6 | 31.5 |
| Released | 2026-08-14 | 2024-09 |
| Weights | Open | Open |
| Context window | 1M | 131K |
| Max output | 131K | 8K |
| Input $ / M tokens | $1.40 | $1.40 |
| Output $ / M tokens | $4.40 | $5.60 |
| Results tracked | 41 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.3 leads
GLM-5.3: 59.0 (#12), Qwen2.5 72B Instruct: 32.8 (#256)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 75.4% | 16% |
| LMArena Coding | 1499 | 1293 |
| DeepSWE | 69% | — |
| FrontierCode | 40.1% | — |
| LMArena WebDev | 1623 | — |
| FrontierSWE | 30.2% | — |
| SciCode | 59% | — |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 1,317 | — |
Agentic & Tool Use GLM-5.3 leads
GLM-5.3: 36.3 (#34), Qwen2.5 72B Instruct: 22.0 (#124)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| APEX-Agents | 56.6% | — |
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
| Vending-Bench 2 | 8,164 | — |
Reasoning GLM-5.3 leads
GLM-5.3: 45.2 (#42), Qwen2.5 72B Instruct: 21.4 (#197)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1491 | 1271 |
| DTBench | 87.7% | 62.9% |
| LMCA | 55.5% | 13.4% |
| Epoch Capabilities Index | 155.56 | 129.01 |
| NYT Connections (extended) | 74.2% | — |
| CritPt | 19.1% | — |
| Chess Puzzles | 21% | — |
| Mystery Game Puzzles | 33% | — |
| Bench to the Future 3 | 0.15 | — |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math GLM-5.3 leads
GLM-5.3: 62.1 (#27), Qwen2.5 72B Instruct: 18.1 (#281)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 91.1% | 8.1% |
| LMArena Math | 1493 | 1283 |
| FrontierMath (Tiers 1-3) | 68.8% | — |
| FrontierMath Tier 4 | 29.3% | — |
| ProofBench | 49% | — |
| Omni-MATH | — | 33% |
| MATH Level 5 | — | 63.2% |
Knowledge GLM-5.3 leads
GLM-5.3: 58.3 (#34), Qwen2.5 72B Instruct: 26.9 (#247)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 90.9% | 49.1% |
| LMArena Expert | 1512 | 1245 |
| SimpleQA Verified | 41% | — |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| GPQA (HELM) | — | 42.6% |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual GLM-5.3 leads
GLM-5.3: 55.9 (#27), Qwen2.5 72B Instruct: 41.2 (#208)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1457 | 1252 |
| LMArena Chinese | 1523 | 1272 |
| LMArena French | 1499 | 1279 |
| LMArena German | 1505 | 1234 |
| LMArena Japanese | 1438 | 1180 |
| LMArena Korean | 1478 | 1188 |
| LMArena Russian | 1460 | 1264 |
| LMArena Spanish | 1444 | 1255 |
Instruction Following GLM-5.3 leads
GLM-5.3: 77.8 (#15), Qwen2.5 72B Instruct: 65.7 (#216)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Instruction Following | 1481 | 1254 |
| IFEval | — | 80.6% |
Long Context GLM-5.3 leads
GLM-5.3: 45.5 (#32), Qwen2.5 72B Instruct: 38.9 (#183)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1485 | 1282 |
Writing & Preference GLM-5.3 leads
GLM-5.3: 75.9 (#6), Qwen2.5 72B Instruct: 46.9 (#210)
| Benchmark | GLM-5.3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1473 | 1269 |
| LMArena Creative Writing | 1459 | 1221 |
| LMArena Multi-Turn | 1473 | 1272 |
| EQ-Bench Creative Writing | 2075 | — |
| WildBench | — | 80.2% |
Frequently asked questions
Is GLM-5.3 better than Qwen2.5 72B Instruct?
GLM-5.3 is the stronger model overall, scoring 54.6 to 31.5 on the Noometry Index.
Which is cheaper, GLM-5.3 or Qwen2.5 72B Instruct?
GLM-5.3 is cheaper. It lists at $1.40 per million input tokens and $4.40 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is GLM-5.3 or Qwen2.5 72B Instruct better for coding?
GLM-5.3 scores higher on coding benchmarks: 59.0 versus 32.8 in the Noometry coding category.
Which has the bigger context window?
GLM-5.3 does, with 1M tokens against 131K.
How many benchmarks do GLM-5.3 and Qwen2.5 72B Instruct share?
23 benchmarks have published results for both models. GLM-5.3 has 41 scored results on Noometry and Qwen2.5 72B Instruct has 43.