Compare / Qwen3.8 Max (0902) vs GLM 5.3
Qwen3.8 Max (0902) vs GLM 5.3
29 results from 4 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Qwen3.8 Max (0902) | GLM 5.3 |
|---|---|---|---|
| ROASBenchmeasured by us | Audience scorescore out of 100 | 39.7 | 51.0 |
| ROASBenchmeasured by us | Business scorescore out of 100 | 15.6 | 12.9 |
| ROASBenchmeasured by us | Consistency scorescore out of 100 | 19.2 | 14.8 |
| ROASBenchmeasured by us | Contribution profitsimulated US dollars | $99,775 | $9,718 |
| ROASBenchmeasured by us | Cost of a runUS dollars | $0.73 | $0.13 |
| ROASBenchmeasured by us | Months over budgetmonths of 12 | 2 | 0 |
| ROASBenchmeasured by us | Overall scorescore out of 100 | 25.4 | 24.8 |
| ROASBenchmeasured by us | Planning scorescore out of 100 | 54.8 | 51.7 |
| ROASBenchmeasured by us | Return on ad spendprofit per $1 spent | $0.09 | $0.01 |
| Arena (formerly LMArena)reported | Confirmed task successIPS effect estimate | 0.05 | 0.06 |
| Arena (formerly LMArena)reported | Praise over complaintIPS effect estimate | 0.04 | 0.06 |
| Arena (formerly LMArena)reported | SteerabilityIPS effect estimate | 0.02 | 0.03 |
| Arena (formerly LMArena)reported | Tool groundingIPS effect estimate | -0.01 | 0.00 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,471 | 1,468 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,467 | 1,456 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,510 | 1,516 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,472 | 1,480 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,479 | 1,480 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,472 | 1,470 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 76.2 | 74.8 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 45.4 | 44.8 |
| Artificial Analysisreported | GPQA Diamond% of questions | 92.8% | 91.7% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 43.1% | 42.3% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 80.3% | 79.7% |
| Artificial Analysisreported | SciCode% of problems | 52.1% | 59.0% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 88.8% | 83.9% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 38.9% | 41.9% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 47.8% | 50.3% |
| SimpleQA Verified (Epoch AI)reported | Correct answers% of questions | 47.3% | 41.0% |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.