Artificial intelligence news.
Straight from the original announcements.

Microsoft's MAI Cyber 1.1 Flash cuts MDASH scan costs by 18%

October 27, 2026, 16:00Source: Microsoft AI
Microsoft's MAI Cyber 1.1 Flash cuts MDASH scan costs by 18%
Image: Microsoft AI

Microsoft AI has released MAI Cyber 1.1 Flash, a vulnerability discovery model that, according to the company, performs comparably to MAI Cyber 1 Flash while costing 18% less per scan. The figure comes from an internal MDASH vulnerability-scanning evaluation suite.

Key points
  • Microsoft AI released MAI Cyber 1.1 Flash for vulnerability discovery inside MDASH.
  • According to Microsoft AI, it performs comparably to MAI Cyber 1 Flash while costing 18% less per scan.
  • The result comes from an internal MDASH vulnerability-scanning evaluation suite.
  • Nearly 9 in 10 scans after a customer's first MAI Cyber 1 Flash scan use the MAI profile, per Microsoft AI.
  • MAI Cyber accounts for almost 50% of all MDASH scans, according to the company.

What changed in MAI Cyber 1.1 Flash

Microsoft AI describes MAI Cyber 1.1 Flash as a step forward in vulnerability discovery. The model builds on MAI Cyber 1 Flash, which the company says showed what is possible when a specialized security model, Microsoft's security data and the MDASH harness work together.

The new version is described as a more focused model that takes on more of the work itself. According to Microsoft AI, this lets the full MDASH system surface useful findings at lower cost.

The headline claim is about efficiency rather than a jump in detection ability. Microsoft AI says the model delivers comparable performance to its predecessor, while each scan costs 18% less to complete. The company bases this on an internal MDASH vulnerability-scanning evaluation suite and an internal product evaluation set, so the result has not been independently verified.

Efficiency across the whole system

Microsoft AI frames the update as more than making a single model cheaper to run. In its words, the goal is bigger: making the entire vulnerability discovery system more efficient while preserving the performance defenders rely on.

The comparison in the announcement is between MAI Cyber 1.1 Flash and MAI Cyber 1 Flash, both running inside MDASH. The 18% lower cost applies to completing the same scans, which Microsoft AI labels as token efficiency.

Customer adoption of MAI Cyber in MDASH

Microsoft AI also shared usage figures for the earlier model. According to the company, after customers run their first scan with the MAI Cyber 1 Flash profile, nearly 9 in 10 subsequent scans use the MAI profile rather than the OpenAI-only model profile.

Microsoft AI says MAI Cyber now accounts for almost 50% of all MDASH scans, with over thousands of scans from hundreds of customers. These are the company's own figures.

What it means for MDASH users

For existing MDASH customers, the update is positioned as a drop-in improvement. Microsoft AI says MAI Cyber 1.1 Flash brings those customers a more focused, efficient model within the workflow they already use.

The source does not give pricing, availability dates or detailed benchmark results beyond the 18% cost figure. Teams that run frequent scans are the ones most likely to notice a per-scan saving, but the size of any real-world benefit will depend on how their workloads compare with Microsoft's internal evaluation set.

Microsoft AIMAI Cyber 1.1 FlashMDASHMAI Cyber 1 Flashcybersecuritylanguage models

Related news

Microsoft uses AI to accelerate Windows vulnerability fixes
Safety and ethicsNovember 5, 2026, 22:09

Microsoft uses AI to accelerate Windows vulnerability fixes

Microsoft is applying AI across the Windows vulnerability management lifecycle, from finding security flaws to validating fixes and monitoring updates after release. The…

CodeRabbit says GPT-6 Cyber increased security findings by 66%
ResearchNovember 5, 2026, 13:00

CodeRabbit says GPT-6 Cyber increased security findings by 66%

CodeRabbit says adding OpenAI’s GPT-6 Cyber to its security scanning ensemble produced 66% more high- and medium-severity findings across 33 benchmark cases. The…

Safety and ethicsTop story
Safety and ethicsNovember 5, 2026, 10:00

OpenAI introduces GPT-6 Cyber for vulnerability research and defense

OpenAI has introduced GPT-6 Cyber, a cybersecurity model designed to find and fix vulnerabilities while more reliably staying within authorized boundaries. The company…