Compare / qwen3-4b-instruct-2507 vs DeepSeek V3.2 Exp
qwen3-4b-instruct-2507 vs DeepSeek V3.2 Exp
11 results from 2 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | qwen3-4b-instruct-2507 | DeepSeek V3.2 Exp |
|---|---|---|---|
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 84.9% | 93.2% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 23.9% | 54.2% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 22.1% | 44.9% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 35.7% | 56.7% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 87.5% | 93.8% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 87.9% | 85.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 76.4% | 76.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 4.5% | 69.5% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 8.0% | 25.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.34 | 0.35 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 29.9 | 54.0 |
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.