Compare / GPT-5 Mini vs grok-4-1-fast-reasoning
GPT-5 Mini vs grok-4-1-fast-reasoning
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 | GPT-5 Mini | grok-4-1-fast-reasoning |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,394 | 1,413 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,381 | 1,417 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,324 | 1,414 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,400 | 1,438 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,373 | 1,401 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,390 | 1,430 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,352 | 1,407 |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 91.0% | 79.4% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 44.3% | 54.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 27.5% | 58.9% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 55.5% | 69.6% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 93.8% | 81.2% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 69.8% | 88.3% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 62.5% | 78.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 82.0% | 82.5% |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 99.9% | 99.7% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 12.9% | 19.2% |
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.