DeepSeek V4 Flash (0420): benchmark results

Provider: DeepSeek. Released 2026-04-24. Access: Open.

Aggregate Rating 1745 ± 30, rank #199 of 10916 rated models, from 51 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA IFBench79.18Accuracy (%)97.6
AA TAU-2 Bench95.61Accuracy (%)95.3
AA GPQA Diamond89.39Accuracy (%)87.2
AA Terminal-Bench Hard38.64Accuracy (%)86
AA Humanity's Last Exam34.85Accuracy (%)80.4
AA CritPt7.14Accuracy (%)73.1
AA Long Context Reasoning74.33Accuracy (%)69.9
Tau3 Banking30.93Success Rate (%)68.9
AA Terminal-Bench 2.161.8Pass@1 (%)60.3
AA GDP.pdf - HR - Criterion Pass Rate70.96Criterion Pass Rate (%)56.9
EnterpriseOps-Gym-AA - Hybrid37.04Task Success (%)53
AA GDP.pdf - Manufacturing/Supply Chains - Criterion Pass Rate62.77Criterion Pass Rate (%)52.2

Interactive version: theaggregate.ai/model?slug=deepseek-v4-flash-0420 · How It Works · Data refreshed daily, snapshot 2026-11-08.