Compare / GPT-4.1 vs GPT-5.2
GPT-4.1 vs GPT-5.2
20 results from 4 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | GPT-4.1 | GPT-5.2 |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,408 | 1,442 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,415 | 1,444 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,402 | 1,391 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,405 | 1,475 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,403 | 1,427 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,415 | 1,437 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,405 | 1,411 |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 86.5% | 87.3% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 23.9% | 45.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 38.9% | 43.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 54.0% | 55.9% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 100.0% | 75.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 88.7% | 81.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 78.9% | 70.4% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 68.0% | 75.5% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 6.0% | 76.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.33 | 0.40 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 67.5 | 49.4 |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 99.9% | 100.0% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 5.6% | 8.4% |
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.