Skip to content

CorpusQA 1M

A million-token CorpusQA long-context question-answering benchmark reported in DeepSeek-V4 model evaluations.

2 models

Higher is better.

  1. 1DeepSeek V4 Pro 0813DeepSeek62.0
  2. 2DeepSeek V4 Flash 0731DeepSeek60.5
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Long-context QA accuracy
Tasks
Million-token corpus question answering
Authors
DeepSeek-AI (2026)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.