From the source
Microsoft unveiled MAI Cyber 2 Flash at Ignite, a specialist security model from
Microsoft AI that the company says delivers leading performance on real-world security tasks at a fraction of the cost of larger models.
- The model is aimed at security teams that need AI to find vulnerabilities, recommend fixes and investigate threats around the clock, a workload where compute bills can quickly add up.
- On CyberGym-E2E, a benchmark that measures real-world vulnerability discovery, MAI Cyber 2 Flash scores 92.3% against 91.5% for GPT-6 Astra, using half the reasoning budget and less than 2% of the cost per task.
- Microsoft calculated that cost at list prices of $0.10 per million input tokens, $0.01 for cached input, and $0.50 for output, compared with $10, $1, and $50 for GPT-6 Astra on Azure
OpenAI Global, and notes the figure reflects offline evaluation cost rather than customer spend.
- At the same reasoning budget, the model also beats GPT-5.6 Cyber on CyberGym-E2E and WebExploitBench, reaching 40.4% against 38.4%.
- Microsoft says it found that 92 of the benchmark's 905 tasks were based on the same open-source projects and vulnerability reports as some of its training data, so it excluded them for every model and reports results on the remaining 813.
- The model is built on Microsoft's decades of proprietary security data and expertise, and brings a one-million-token context window, four times larger than MAI Cyber 1.1 Flash, so it can analyze more of a codebase during complex, multi-step investigations.
- MAI Cyber 2 Flash is coming exclusively to Microsoft security products, including Codename MDASH, where it will automatically replace the existing model in the MAI profile for eligible customers starting in mid-December.