Microsoft AI's real-time speech-to-text model for voice agents, transcribing 60 languages with continuous language detection
MAI Cyber 2 Flash
New· 4Microsoft's specialist security model announced 17 November 2026: one-million-token context, list price $0.10 input, $0.01 cached input and $0.50 output per million tokens, coming only inside Microsoft security products from mid-December.
TrustList summary
Not yet independently verified. Only Microsoft's own announcement page and one tracker, a Product Hunt entry dated 17 November 2026, back this model. The benchmark results, the cost comparison and the customer figure are Microsoft's own and have not been checked independently, so no scores are stored here. The page does not give a licence or say where the model can be called outside Microsoft's own security products. We will update this when it can be confirmed, and remove this note.
Microsoft announced MAI Cyber 2 Flash on 17 November 2026 at its Ignite event. It is a specialist security model, built on Microsoft's own security data, meant to find vulnerabilities, recommend fixes and investigate threats at lower cost than larger general models.
Where it will run
The model is not sold as a general API. Microsoft says it is coming exclusively to its own security products, including the one it calls Codename MDASH, where it will automatically replace the existing model in the MAI profile for eligible customers from mid-December 2026.
Context and price
The context window is one million tokens, four times that of MAI Cyber 1.1 Flash, so it can take in more of a codebase during multi-step investigations. In the footnotes to its cost comparison Microsoft gives list prices of $0.10, $0.01 and $0.50 per million tokens for input, cached input and output. It lists a larger rival model on Azure at $10, $1 and $50. Microsoft's cost figures describe offline evaluation cost, not what a customer pays.
What Microsoft reports
Microsoft's own results: on CyberGym-E2E, a benchmark of real-world vulnerability discovery, it reports 92.3% task success against 91.5% for a larger rival model using half the reasoning budget, at under 2% of the cost per task. It excluded 92 of the 905 tasks because they overlap with its training data, and reports on the other 813. On WebExploitBench it reports 40.4% against 38.4% for another security model at the same reasoning budget. A comparison with a further security model is promised in the coming weeks. One customer reports its own evaluations showed 20% better token efficiency than several older rival models. These figures are Microsoft's and are not treated as scores here.
Key facts
Maker: Microsoft Announced: 17 November 2026 (Ignite) Context window: 1,000,000 tokens Price per million tokens: $0.10 input, $0.01 cached input, $0.50 output (list, per Microsoft) Access: Microsoft security products only; from mid-December 2026 in Codename MDASH Licence: not stated; hosted by Microsoft
Benchmarks & AI stats
| Benchmark | Official | Community avg |
|---|---|---|
| No benchmark scores yet. Be the first to add one. | ||
“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).
Sign in to add a benchmark score for this model.
MAI Cyber 2 Flash launches
Trust Score
An earned signal from verification, reviews, awards, transparency and engagement - the vendor can't buy it.
Updated 11/18/2026
Request a demo or quote from MAI Cyber 2 Flash
Reviews
Write the first review of MAI Cyber 2 Flash
Used it? Your experience helps other buyers decide.
Write a reviewQuestions & answers
No questions yet. Be the first to ask about MAI Cyber 2 Flash.
MAI Cyber 2 Flash alternatives
Similar listings buyers compare against this one.
Google's open multimodal embedding model that maps text, images, audio and video into one vector space, under Apache 2.0
Anthropic's smallest and fastest Claude 5.5 model, available on its own API and three cloud platforms
JetBrains' open 12B mixture-of-experts coding model with 2.5B parameters active per token under Apache 2.0
Aleph Alpha's open-weight English-German mixture-of-experts model for self-hosting, released under Apache 2.0 in October 2026
Microsoft AI's high-volume, low-latency variant of MAI-Voice-2.1 for speech in 23 languages, aimed at voice agents
Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems
Black Forest Labs' image generation and editing model with bounding-box layouts, multiple reference images and 4K output
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- Write a reviewHelp other buyers decide
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- ArticlesGuides, comparisons and research from the editorial desk and the community.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.