Compare / gpt-oss-20b vs GLM 5.1
gpt-oss-20b vs GLM 5.1
21 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | gpt-oss-20b | GLM 5.1 |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,323 | 1,465 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,240 | 1,448 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,314 | 1,495 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,281 | 1,460 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,318 | 1,465 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,275 | 1,453 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 20.7 | 55.8 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 10.0 | 26.1 |
| Artificial Analysisreported | GPQA Diamond% of questions | 68.8% | 86.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 11.0% | 30.1% |
| Artificial Analysisreported | IFBench% of instructions | 65.1% | 76.3% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 34.7% | 73.7% |
| Artificial Analysisreported | SciCode% of problems | 38.9% | 44.8% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 13.9% | 61.8% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 0.0% | 2.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 10.6% | 43.2% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 7.0% | 13.6% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 60.2% | 97.7% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 98.0% | 9.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.31 | 0.38 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 24.6 | 61.7 |
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.