Compare / Qwen3 235B A22B Instruct 2507 vs GPT-5 Mini
Qwen3 235B A22B Instruct 2507 vs GPT-5 Mini
22 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Qwen3 235B A22B Instruct 2507 | GPT-5 Mini |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,431 | 1,394 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,432 | 1,381 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,378 | 1,324 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,448 | 1,400 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,414 | 1,373 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,422 | 1,390 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,396 | 1,352 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 12.0 | 20.6 |
| Artificial Analysisreported | GPQA Diamond% of questions | 75.3% | 82.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 11.1% | 21.5% |
| Artificial Analysisreported | IFBench% of instructions | 46.1% | 75.4% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 33.9% | 72.3% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 15.2% | 33.3% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 33.3% | 71.1% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 81.7% | 91.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 23.9% | 44.3% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 45.4% | 27.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 52.1% | 55.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 93.8% | 93.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 90.3% | 69.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 78.7% | 62.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 54.0% | 82.0% |
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.