Compare / DeepSeek V3.1 vs Gemini 3.1 Flash Lite Preview
DeepSeek V3.1 vs Gemini 3.1 Flash Lite Preview
15 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 V3.1 | Gemini 3.1 Flash Lite Preview |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,417 | 1,421 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,402 | 1,413 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,432 | 1,443 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,416 | 1,407 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,417 | 1,433 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,407 | 1,420 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 13.7 | 15.6 |
| Artificial Analysisreported | GPQA Diamond% of questions | 77.9% | 82.2% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 14.3% | 17.2% |
| Artificial Analysisreported | IFBench% of instructions | 41.5% | 77.2% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 56.7% | 74.3% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 25.0% | 24.2% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 37.4% | 31.3% |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 94.5% | 99.6% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 5.5% | 8.2% |
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.