Compare / gpt-oss-120b vs gpt-oss-20b
gpt-oss-120b vs gpt-oss-20b
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-120b | gpt-oss-20b |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,351 | 1,323 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,277 | 1,240 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,357 | 1,314 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,323 | 1,281 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,352 | 1,318 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,309 | 1,275 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 30.4 | 20.7 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 11.6 | 10.0 |
| Artificial Analysisreported | GPQA Diamond% of questions | 78.2% | 68.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 19.6% | 11.0% |
| Artificial Analysisreported | IFBench% of instructions | 69.0% | 65.1% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 52.0% | 34.7% |
| Artificial Analysisreported | SciCode% of problems | 34.0% | 38.9% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 26.2% | 13.9% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 0.0% | 0.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 23.5% | 10.6% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 12.8% | 7.0% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 65.8% | 60.2% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 107.0% | 98.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.29 | 0.31 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 38.5 | 24.6 |
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.