Compare / Nova Pro 1.0 vs Gemma 3 12B
Nova Pro 1.0 vs Gemma 3 12B
16 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Nova Pro 1.0 | Gemma 3 12B |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,279 | 1,386 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,237 | 1,332 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,273 | 1,274 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,277 | 1,321 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,290 | 1,342 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,274 | 1,313 |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 70.1% | 70.3% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 1.9% | 27.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 1.9% | 5.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 25.0% | 30.4% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 93.8% | 93.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 86.6% | 79.4% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 78.5% | 74.2% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 2.5% | 4.0% |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 99.3% | 97.4% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 5.1% | 4.4% |
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.