DeepDi: A Fast and Accurate Disassembler for Binary Code AI

AI models (e.g., malware detection) work perfectly, but feature extraction (e.g., n-gram, CFG) from binary code is too slow?

Built for Efficient Binary Code Analysis

  • GPU-Accelerated


    Unleash the power of GPU for disassembly, deliberately built for downstream machine learning based cybersecurity applications (e.g., malware detection and classification).

  • Obfuscation Resilient

    Obfuscation Resilient

    Low false positive/negative rate on identifying obfuscated code, and can automatically handle unseen obfuscation techniques with quick retraining.

  • Accurate & Efficient

    Accurate & Efficient

    Disassemble the code 350X faster than state-of-art disassemblers while producing the same or better accuracy.

  • 350X

    Faster than state-of-the-art

  • 0.02%/0.02%

    False Positive/Negative Rate on instruction identification

  • 99.9%

    Precision on Function Identification


  • Save Manual Analysis Time

    Save Manual Analysis Time

    Disassemble those large binary files that take other disassemblers hours to analyze within seconds.

  • Analyze Code on Endpoint in Real Time

    Analyze Code on Endpoint in Real Time

    Embedded with endpoint security solutions to identify the malware promptly.

  • Empower Large Volume Code Analysis

    Empower Large Volume Code Analysis

    Conduct sophisticated analysis of large volume code that cannot be finished timely before.

  • Enable Memory Dump Analysis of Cloud VMs

    Enable Memory Dump Analysis of Cloud VMs

    Dump and analyze the memory of Cloud VMs periodically to find any in-memory threats timely.



X86/64/ARM support


Docker container support


Linux support


GPU support


Powerful APIs

Get DeepDi Community Edition
For Free