arcanicai/Con0

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 18, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Con0 is a 27 billion parameter vision-language model developed by Arcanic AI and NextZero, featuring a hybrid Gated DeltaNet and full attention architecture. Specialized for cybersecurity, it analyzes code, logs, screenshots, and infrastructure diagrams to identify vulnerabilities and provide remediation guidance. With a 262,144-token context window, Con0 excels at auditing web, application, AI, and general security aspects, including SQL injection, XSS, prompt injection, and incident analysis.

Loading preview...

Con0: A Vision-Language Model for Cybersecurity Auditing

Con0 is a 27 billion parameter vision-language model (VLM) developed by Arcanic AI in collaboration with NextZero, specifically designed to function as a security auditor. It integrates a hybrid Gated DeltaNet and full attention architecture, enabling it to process and understand diverse inputs including images, video, and OCR, alongside text.

Key Capabilities

Con0 is engineered to identify and surface vulnerabilities across various security domains, providing actionable remediation guidance. Its core capabilities include:

  • Web & Application Security: Detects common flaws like SQL injection (SQLi), Cross-site scripting (XSS), CSRF, SSRF, command injection, path traversal, XXE, deserialization, and issues related to authentication and access control. It covers the OWASP Top 10 and assists with secure-configuration assessment.
  • AI & Agent Security: Analyzes prompt injection and jailbreak surfaces, assesses tool-calling and function-dispatch misuse, and reviews least-privilege/permission scopes for multi-step agent threat modeling.
  • General Security: Facilitates vulnerability triage, CVE context mapping, and incident analysis from logs, traces, and screenshots, offering secure-code guidance.

Model Architecture & Context

This VLM features a substantial 262,144-token context window, allowing for comprehensive analysis of large codebases, extensive logs, and complex diagrams. It also incorporates flexible reasoning control and tool-calling functionalities.

Good For

Con0 is ideal for developers and security professionals who need an AI assistant capable of:

  • Automated security auditing of code and infrastructure.
  • Identifying and mitigating web, application, and AI-specific vulnerabilities.
  • Analyzing security incidents from multimodal data (logs, screenshots).
  • Generating secure-code guidance and remediation plans.

Finetunes
2 models