Compare / gpt-oss-120b vs Grok 4.6
gpt-oss-120b vs Grok 4.6
21 results from 5 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | gpt-oss-120b | Grok 4.6 |
|---|---|---|---|
| APEX-Agentsreported | Tasks passed% of tasks | 4.4% | 65.3% |
| Arena (formerly LMArena)reported | OverallArena rating | 1,385 | 1,449 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,351 | 1,447 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,277 | 1,445 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,357 | 1,490 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,323 | 1,451 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,352 | 1,453 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,309 | 1,446 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 30.4 | 76.8 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 11.6 | 44.3 |
| Artificial Analysisreported | GPQA Diamond% of questions | 78.2% | 94.9% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 19.6% | 44.1% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 52.0% | 81.0% |
| Artificial Analysisreported | SciCode% of problems | 34.0% | 56.5% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 26.2% | 88.4% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 0.0% | 21.2% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 12.8% | 50.7% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 107.0% | 25.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.29 | 0.36 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 38.5 | 63.3 |
| Vending-Bench 2reported | Money after a yearUS dollars | $-21.53 | $9,047.03 |
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.