Compare / Command A vs Phi 4
Command A vs Phi 4
19 results from 4 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Command A | Phi 4 |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,351 | 1,256 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,336 | 1,210 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,344 | 1,268 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,342 | 1,245 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,354 | 1,256 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,348 | 1,229 |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 84.2% | 87.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 16.6% | 24.7% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 29.5% | 3.9% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 46.5% | 28.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 81.2% | 50.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 87.6% | 69.6% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 78.5% | 60.7% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 46.5% | 4.5% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 21.0% | 19.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.35 | 0.34 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 40.6 | 25.7 |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 97.6% | 80.7% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 9.3% | 3.7% |
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.