Compare / Claude Sonnet 5.5 vs DeepSeek V4 Pro 0423
Claude Sonnet 5.5 vs DeepSeek V4 Pro 0423
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 | Claude Sonnet 5.5 | DeepSeek V4 Pro 0423 |
|---|---|---|---|
| ROASBenchmeasured by us | Audience scorescore out of 100 | 61.8 | 44.8 |
| ROASBenchmeasured by us | Business scorescore out of 100 | 57.2 | 23.8 |
| ROASBenchmeasured by us | Consistency scorescore out of 100 | 28.5 | 9.1 |
| ROASBenchmeasured by us | Contribution profitsimulated US dollars | $571,203 | $301,289 |
| ROASBenchmeasured by us | Cost of a runUS dollars | $0.42 | $0.20 |
| ROASBenchmeasured by us | Months over budgetmonths of 12 | 0 | 0 |
| ROASBenchmeasured by us | Overall scorescore out of 100 | 52.0 | 28.2 |
| ROASBenchmeasured by us | Planning scorescore out of 100 | 55.1 | 54.0 |
| ROASBenchmeasured by us | Return on ad spendprofit per $1 spent | $0.70 | $0.26 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 56.0 | 36.0 |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 55.0% | 41.0% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 82.7% | 80.3% |
| Artificial Analysisreported | SciCode% of problems | 61.0% | 51.0% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 63.6% | 14.1% |
| SimpleQA Verified (Epoch AI)reported | Correct answers% of questions | 46.5% | 47.0% |
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.