DeepSeek-V4-Flash-FP8 — benchmark results
DeepSeek V4 Flash evaluated in FP8 quantization. Provider: DeepSeek. Released 2026-04-23. Access: Open.
Unified ELO 1725 ± 22, rank #227 of 1776 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| RewardBench 2 Focus | 93.64 | Accuracy (%) | 100 |
| GSM-MC | 99.47 | Accuracy (%) | 97.8 |
| MATH-MC Level 5 | 99.53 | Accuracy (%) | 78.7 |
| RewardBench 2 Precise IF | 68.91 | Accuracy (%) | 76.5 |
| MATH-MC Level 4 | 99.33 | Accuracy (%) | 75 |
| JudgeBench Coding | 96.43 | Accuracy (%) | 73.5 |
| RewardBench 2 Safety | 87.56 | Accuracy (%) | 68.6 |
| MATH-MC Level 3 | 98.92 | Accuracy (%) | 65.4 |
| RewardBench 2 Math | 88.93 | Accuracy (%) | 63.7 |
| JudgeBench Math | 89.29 | Accuracy (%) | 60.8 |
| RewardBench 2 Factuality | 76.68 | Accuracy (%) | 58.8 |
| JudgeBench Knowledge | 77.6 | Accuracy (%) | 56.9 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-flash-fp8 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.