Compare / Llama 3.2 3B Instruct vs o3

Llama 3.2 3B Instruct vs o3

17 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.

Llama 3.2 3B Instruct
Meta · profile
o3
OpenAI · profile
Results compared
17
Shared sources
3
BenchmarkMetricLlama 3.2 3B Instructo3
Arena (formerly LMArena)reportedBusiness, management and financeArena rating 1,1461,426
Arena (formerly LMArena)reportedCreative writingArena rating 1,1451,383
Arena (formerly LMArena)reportedExpert promptsArena rating 1,1791,449
Arena (formerly LMArena)reportedInstruction followingArena rating 1,1461,403
Arena (formerly LMArena)reportedOverallArena rating 1,1671,432
Arena (formerly LMArena)reportedWriting, literature and languageArena rating 1,1431,396
Berkeley Function Calling Leaderboard (BFCL) V4reportedIrrelevance detection% correct 52.1%86.1%
Berkeley Function Calling Leaderboard (BFCL) V4reportedMemory% correct 6.2%51.8%
Berkeley Function Calling Leaderboard (BFCL) V4reportedMulti-turn tasks% correct 4.0%62.2%
Berkeley Function Calling Leaderboard (BFCL) V4reportedOverall accuracy% correct 21.9%63.0%
Berkeley Function Calling Leaderboard (BFCL) V4reportedRelevance detection% correct 87.5%93.8%
Berkeley Function Calling Leaderboard (BFCL) V4reportedSingle-turn calls (curated)% correct 82.7%81.9%
Berkeley Function Calling Leaderboard (BFCL) V4reportedSingle-turn calls (user-contributed)% correct 58.3%73.2%
Berkeley Function Calling Leaderboard (BFCL) V4reportedWeb search% correct 1.0%77.0%
UGI LeaderboardreportedRequested-length error% off the requested word count 31.0%18.0%
UGI LeaderboardreportedStyle adherencescore from 0 to 1 0.360.30
UGI LeaderboardreportedWriting scorescore out of 100 10.466.8

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.