Compare / DeepSeek V3.2 vs grok-4-fast-non-reasoning
DeepSeek V3.2 vs grok-4-fast-non-reasoning
12 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | DeepSeek V3.2 | grok-4-fast-non-reasoning |
|---|---|---|---|
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 21.5 | 11.1 |
| Artificial Analysisreported | GPQA Diamond% of questions | 84.0% | 60.6% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 24.6% | 4.5% |
| Artificial Analysisreported | IFBench% of instructions | 60.7% | 37.7% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 73.3% | 24.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 35.6% | 12.1% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 90.6% | 63.7% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 26.0% | 19.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.36 | 0.33 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 57.3 | 43.0 |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 92.6% | 99.2% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 6.3% | 19.7% |
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.