Compare / Llama 3.2 3B Instruct vs Kimi K2 0711
Llama 3.2 3B Instruct vs Kimi K2 0711
17 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Llama 3.2 3B Instruct | Kimi K2 0711 |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,146 | 1,409 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,145 | 1,371 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,179 | 1,417 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,146 | 1,381 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,167 | 1,418 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,143 | 1,375 |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 52.1% | 87.3% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 6.2% | 29.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 4.0% | 50.6% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 21.9% | 59.1% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 87.5% | 75.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 82.7% | 81.6% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 58.3% | 78.7% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 1.0% | 66.5% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 31.0% | 22.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.36 | 0.29 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 10.4 | 52.5 |
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.