DeepSeek V4 Flash (Preview): benchmark results
Provider: DeepSeek. Released 2026-04-24.
Aggregate Rating 1699 ± 27, rank #354 of 10916 rated models, from 23 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| GigaChat Reasoning Card - Function Calling V4 | 68.06 | Score (%) | 100 |
| GigaChat Reasoning Card - GPQA-Diamond | 87.4 | Score (%) | 100 |
| GigaChat Reasoning Card - HMMT 2025, mean@8 | 95.21 | Score (%) | 100 |
| GigaChat Reasoning Card - IMOAnswerBench | 85.75 | Score (%) | 100 |
| GigaChat Reasoning Card - Live Code Bench v6 | 87.87 | Score (%) | 100 |
| GigaChat Reasoning Card - Natural Plan | 88 | Score (%) | 100 |
| GigaChat Reasoning Card - SWE-bench Verified | 78.6 | Score (%) | 100 |
| GigaChat Reasoning Card - TAU3-bench | 67.7 | Score (%) | 100 |
| GigaChat Reasoning Card - Terminal-Bench 2 | 56.6 | Score (%) | 100 |
| LLM2014 Logic 2026-07 | 36.24 | Median Score | 58.1 |
| GigaChat Reasoning Card - AIME 2025, mean@32 | 88.95 | Score (%) | 50 |
| GigaChat Reasoning Card - AIME 2026, mean@32 | 90.4 | Score (%) | 50 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-flash-preview · How It Works · Data refreshed daily, snapshot 2026-11-08.