Release 7.14.0
Corellium 7.14 introduces comprehensive support for iOS adn iPadOS 27.0 following its September GA release. This update also delivers AI agent capabilities via Corellium-developed MCP servers, available as both a fully hosted service and a self-hosted local option. Furthermore, This release advances our ongoing API/SDK modernization, expands Mirror availability as a purchasable add-on for forensic examiners, and broadens Apple platform support across the iOS/iPadOS 18, 26, and 27 lines. This release also adds iPhone 18 device support, with Falcon Premium customers gaining early access to iOS/iPadOS 27.x betas.
Mobile OS Support
We've expanded out platform's compatibility with the latest mobile operating system versions:
New Apple Device Support
- Added Support for iPhone 18
Apple Platform Support
- Added support for iOS/iPadOS 27.2.0 beta 1 and beta 2 (for SA Falcon Premium only)
- Added support for iOS/iPadOS 27.0.1
- Added support for iOS/iPadOS 27.0.0 betas 5-8 (for SA Falcon Premium only)
- Added support for iOS/iPadOS 26.7.0
- Added support for iOS/iPadOS 26.6.2
- Added support for iOS/iPadOS 26.6.1
- Added support for iOS/iPadOS 26.6.0 RC1/GA
- Added support for iOS/iPadOS 18.7.10
Early access to beta operating systems remains available for Falcon Premium customers, enabling security teams, developers, and researchers to validate applications and workflows against upcoming Apple releases before public availability.
New Features
Corellium API/SDK/MCP
Corellium's API and MCP tooling continue a multi-release modernization effort (started in 7.12 and continued through 7.13), unifying field naming across the REST API, publishing a versioned OpenAPI spec, and generating SDKs directly from it. In 7.14, this work extends to a fully cluster-hosted MCP server, giving every customer, cloud or self-hosted, AI agent access with no setup required, plus a round of hardening and bug fixes:
-
MCP Server now cluster-hosted. The Corellium MCP server runs directly on Corellium infrastructure. Cloud customers get MCP access the moment they log in, with no infrastructure to manage; self-hosted (SA/DA) customers get it bundled with their existing deployment. Authentication uses existing Corellium credentials, with full tenant isolation between customers.
-
MCP reliability and security hardening, including: fixes for upload JSON parsing failures and repeated upload loops, corrected checksum guidance for extension uploads, credential redaction from default conversational MCP results, support for larger HTTP file uploads, and a rate limiter keyed by authenticated caller rather than proxy address.
-
Frontend SDK adoption continues, with the Corellium web app progressively replacing direct HTTP calls with the generated TypeScript SDK (session tokens/auth, image uploads, and more), plus a bug-bash pass addressing API field-naming and schema issues surfaced during adoption.
-
API docs rebuild, moving
/api/docsto a new UI built on Scalar, with ongoing investigation into surfacing SDK method names and per-language type signatures directly in the docs.
Benefit: AI agents and automation tooling get a single, typed, consistently-shaped way to interact with Corellium regardless of deployment model, and the web app itself is migrating onto the same modern API surface.
Mirror
Mirror is now publicly available for purchase as an add-on to Falcon Essentials or Falcon Premium, and can also be trialed. Mirror takes a Cellebrite UFED (UFD) extraction and loads it directly into a new Corellium iOS virtual device, so the extracted data can be reviewed by navigating the mobile app’s pre-seeded user data.
Benefit: Examiners can investigate an extraction the way a suspect actually experienced it, opening apps, scrolling conversations, and navigating native interfaces, rather than piecing together context from a parsed report, leading to faster and more intuitive investigations.
Other Enhancements
Frida Framework Updates
Updated platform integration to support Frida 17.19.0 for iOS, and 17.18.0 for Android.
Bug Fixes & UI Enhancements
Various underlying bugs have been resolved to improve stability and performance. Minor user interface tweaks have been implemented for a more intuitive and user-friendly experience.



We made a stack of iterative improvements to our MATRIX feature!


