Post-trained update to DeepSeek V4 Flash with stronger reasoning, coding, and agentic performance.
SI v3 uses weighted percentages from one Artificial Analysis configuration. Ranking requires all five benchmarks. Partial scores average only available weights. Scores are not comparable to SI v2 or a probability of reaching singularity.
Compared with models released within six months of it.
| Benchmark | Score | Rank |
|---|---|---|
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 91.3% | #31 / 96 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 34.5% | #59 / 102 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisDeepSeek V4 Flash Vision (Max) | 34.5% |
MMMU-Pro Artificial AnalysisDeepSeek V4 Flash Vision (Max) | 74.8% |
Terminal-Bench 2.1 Artificial AnalysisDeepSeek V4 Flash Vision (Max) | 74.2% |
Terminal-Bench 4.0 Artificial AnalysisDeepSeek V4 Flash Vision (Max) | 12.1% |
SciCode Artificial AnalysisDeepSeek V4 Flash Vision (Max) | 49.7% |
AA-LCR Artificial AnalysisDeepSeek V4 Flash Vision (Max) | 81.3% |
CritPt Artificial AnalysisDeepSeek V4 Flash Vision (Max) | 10.9% |