Compare / DeepSeek V4 Pro 0423 vs Mistral Large 4
DeepSeek V4 Pro 0423 vs Mistral Large 4
32 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 V4 Pro 0423 | Mistral Large 4 |
|---|---|---|---|
| DeckBenchmeasured by us | Accurate decks% of tasks | 0.0% | 0.0% |
| DeckBenchmeasured by us | Caveat dropped% of tasks | 16.7% | 16.7% |
| DeckBenchmeasured by us | Clean layout% of tasks | 33.3% | 33.3% |
| DeckBenchmeasured by us | Cost per deckUS dollars | $0.0465 | $0.0148 |
| DeckBenchmeasured by us | Deck ratingrating | 615 | 515 |
| DeckBenchmeasured by us | Design quality% of the maximum | 40.9% | 30.5% |
| DeckBenchmeasured by us | Draft figure quoted% of tasks | 33.3% | 16.7% |
| DeckBenchmeasured by us | Findings missing% of tasks | 16.7% | 0.0% |
| DeckBenchmeasured by us | Head-to-head win rate% of comparisons | 18.2% | 10.7% |
| DeckBenchmeasured by us | Layout defects% of tasks | 66.7% | 66.7% |
| DeckBenchmeasured by us | Misleading metric used% of tasks | 0.0% | 0.0% |
| DeckBenchmeasured by us | Presentable decks% of tasks | 0.0% | 0.0% |
| DeckBenchmeasured by us | Recommendation late or wrong% of tasks | 66.7% | 83.3% |
| DeckBenchmeasured by us | Slides needing work% of tasks | 83.3% | 100.0% |
| DeckBenchmeasured by us | Unsupported claims% of tasks | 66.7% | 50.0% |
| DeckBenchmeasured by us | Unsupported numbers% of tasks | 0.0% | 33.3% |
| ROASBenchmeasured by us | Audience scorescore out of 100 | 44.8 | 55.8 |
| ROASBenchmeasured by us | Business scorescore out of 100 | 23.8 | 34.2 |
| ROASBenchmeasured by us | Consistency scorescore out of 100 | 9.1 | 16.7 |
| ROASBenchmeasured by us | Contribution profitsimulated US dollars | $301,289 | $533,401 |
| ROASBenchmeasured by us | Cost of a runUS dollars | $0.20 | $0.0657 |
| ROASBenchmeasured by us | Months over budgetmonths of 12 | 0 | 0 |
| ROASBenchmeasured by us | Overall scorescore out of 100 | 28.2 | 37.0 |
| ROASBenchmeasured by us | Planning scorescore out of 100 | 54.0 | 55.0 |
| ROASBenchmeasured by us | Return on ad spendprofit per $1 spent | $0.26 | $0.45 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 36.0 | 38.4 |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 41.0% | 35.0% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 80.3% | 81.3% |
| Artificial Analysisreported | Output speedtokens per second | 117 | 106 |
| Artificial Analysisreported | SciCode% of problems | 51.0% | 54.2% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 14.1% | 26.8% |
| Artificial Analysisreported | Time to first answer tokenseconds | 0.90 | 19.8 |
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.