Compare / Gemma 4 26B A4B vs GPT-5.5
Gemma 4 26B A4B vs GPT-5.5
23 results from 4 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Gemma 4 26B A4B | GPT-5.5 |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,445 | 1,486 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,444 | 1,489 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,399 | 1,452 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,475 | 1,513 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,437 | 1,478 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,438 | 1,481 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,406 | 1,470 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 39.3 | 74.9 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 16.7 | 38.4 |
| Artificial Analysisreported | GPQA Diamond% of questions | 79.2% | 93.5% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 19.3% | 45.8% |
| Artificial Analysisreported | IFBench% of instructions | 72.4% | 75.8% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 65.7% | 84.3% |
| Artificial Analysisreported | SciCode% of problems | 40.0% | 56.1% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 39.0% | 84.3% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 25.0% | 60.6% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 12.0% | 39.0% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 43.6% | 93.9% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 8.0% | 13.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.35 | 0.41 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 43.8 | 69.7 |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 99.8% | 100.0% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 5.2% | 9.3% |
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.