Model comparison
GLM-5.3 vs Qwen3.8 27B
GLM-5.3 is the stronger model overall, scoring 54.6 to 45.3 on the Noometry Index. Qwen3.8 27B costs 1.9× less per token, which makes it the better buy when GLM-5.3's lead doesn't matter for your workload.
Last verified . 26 shared benchmarks.
Alibaba (Qwen)Summary
- They share 26 benchmarks with published results for both. GLM-5.3 scores higher in 8 categories and Qwen3.8 27B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.3 leads 62.1 to 35.7.
- The biggest single-benchmark swing is ProofBench: 49% for GLM-5.3 and 16% for Qwen3.8 27B.
- Qwen3.8 27B is cheaper at $0.99 / $1.49 per million input/output tokens, against $1.40 / $4.40 for GLM-5.3.
- GLM-5.3 accepts more context: 1M tokens versus 262K.
Side by side
| GLM-5.3 | Qwen3.8 27B | |
|---|---|---|
| Provider | Z.ai (Zhipu) | Alibaba (Qwen) |
| Noometry Index | 54.6 | 45.3 |
| Released | 2026-08-14 | 2026-08-14 |
| Weights | Open | Open |
| Context window | 1M | 262K |
| Max output | 131K | 33K |
| Input $ / M tokens | $1.40 | $0.99 |
| Output $ / M tokens | $4.40 | $1.49 |
| Results tracked | 41 | 28 |
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), Qwen3.8 27B: 50.3 (#42)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| LMArena WebDev | 1623 | 1592 |
| SciCode | 59% | 46.6% |
| LMArena Coding | 1499 | 1487 |
| DeepSWE | 69% | — |
| FrontierCode | 40.1% | — |
| FrontierSWE | 30.2% | — |
| WeirdML | 75.4% | — |
| ALE-Bench | 1,317 | — |
Agentic & Tool Use Not comparable
GLM-5.3: 36.3 (#34), Qwen3.8 27B: —
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| APEX-Agents | 56.6% | — |
| Vending-Bench 2 | 8,164 | — |
Reasoning GLM-5.3 leads
GLM-5.3: 45.2 (#42), Qwen3.8 27B: 31.5 (#71)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| NYT Connections (extended) | 74.2% | 54.5% |
| CritPt | 19.1% | 5.4% |
| LMArena Hard Prompts | 1491 | 1459 |
| DTBench | 87.7% | 88% |
| LMCA | 55.5% | 41.4% |
| Epoch Capabilities Index | 155.56 | 149.38 |
| Chess Puzzles | 21% | — |
| Mystery Game Puzzles | 33% | — |
| Surface Evolver Bench | — | 45% |
| Bench to the Future 3 | 0.15 | — |
Math GLM-5.3 leads
GLM-5.3: 62.1 (#27), Qwen3.8 27B: 35.7 (#139)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| ProofBench | 49% | 16% |
| LMArena Math | 1493 | 1453 |
| FrontierMath (Tiers 1-3) | 68.8% | — |
| FrontierMath Tier 4 | 29.3% | — |
| OTIS Mock AIME 2024-2025 | 91.1% | — |
Knowledge GLM-5.3 leads
GLM-5.3: 58.3 (#34), Qwen3.8 27B: 40.9 (#105)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1512 | 1482 |
| GPQA Diamond | 90.9% | — |
| SimpleQA Verified | 41% | — |
Multimodal Not comparable
GLM-5.3: —, Qwen3.8 27B: 40.5 (#36)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | — | 1270 |
Multilingual GLM-5.3 leads
GLM-5.3: 55.9 (#27), Qwen3.8 27B: 54.0 (#54)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1457 | 1431 |
| LMArena Chinese | 1523 | 1504 |
| LMArena French | 1499 | 1468 |
| LMArena German | 1505 | 1435 |
| LMArena Japanese | 1438 | 1392 |
| LMArena Korean | 1478 | 1389 |
| LMArena Russian | 1460 | 1421 |
| LMArena Spanish | 1444 | 1444 |
Instruction Following GLM-5.3 leads
GLM-5.3: 77.8 (#15), Qwen3.8 27B: 75.8 (#54)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1481 | 1436 |
Long Context GLM-5.3 leads
GLM-5.3: 45.5 (#32), Qwen3.8 27B: 44.3 (#66)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1485 | 1448 |
Writing & Preference GLM-5.3 leads
GLM-5.3: 75.9 (#6), Qwen3.8 27B: 65.9 (#41)
| Benchmark | GLM-5.3 | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1473 | 1441 |
| LMArena Creative Writing | 1459 | 1382 |
| EQ-Bench Creative Writing | 2075 | 1671 |
| LMArena Multi-Turn | 1473 | 1443 |
Frequently asked questions
Is GLM-5.3 better than Qwen3.8 27B?
GLM-5.3 is the stronger model overall, scoring 54.6 to 45.3 on the Noometry Index. Qwen3.8 27B costs 1.9× less per token, which makes it the better buy when GLM-5.3's lead doesn't matter for your workload.
Which is cheaper, GLM-5.3 or Qwen3.8 27B?
Qwen3.8 27B is cheaper. It lists at $0.99 per million input tokens and $1.49 per million output tokens; GLM-5.3 lists at $1.40 and $4.40.
Is GLM-5.3 or Qwen3.8 27B better for coding?
GLM-5.3 scores higher on coding benchmarks: 59.0 versus 50.3 in the Noometry coding category.
Which has the bigger context window?
GLM-5.3 does, with 1M tokens against 262K.
How many benchmarks do GLM-5.3 and Qwen3.8 27B share?
26 benchmarks have published results for both models. GLM-5.3 has 41 scored results on Noometry and Qwen3.8 27B has 28.