Ubuntu 26.10 is here with Linux 7.3 kernel, GNOME 51, Rust coreutils & Myna speech-to-text

Canonical has announced that Ubuntu 26.10, code-named “Stonking Stingray”, is now available for download and installation. The company framed the release as one that both consolidates earlier groundwork and reaches toward newer territory, highlighting completed migrations, expanded hardware support, and the arrival of Ubuntu's first AI feature.

On the software front, Ubuntu 26.10 ships with GNOME 51, delivering a smoother, more responsive desktop with better display handling and improved accessibility. The release also includes the Linux kernel 7.3, which enables hardware support for the latest silicon including Intel Xe3P (Nova Lake), AMD Zen 6, and NVIDIA 615 processors. The kernel brings enhanced direct rendering management and power management hooks for heterogeneous platforms spanning CPU cores, GPUs, and media decoders, plus out-of-the-box support for Lustre, a parallel distributed file system for transferring massive datasets across corporate nodes.

On the security side, Canonical said the release completes the migration to Rust-based uutils coreutils and introduces post-quantum-ready package signing by replacing GnuPG with Sequoia PGP to verify APT repository signatures. Ubuntu 26.10 also adds quantum-safe token support via openCryptoki with PKCS #11 V3.2 coverage across Soft, EP11, and CCA tokens integrating ML-DSA and ML-KEM, and it broadens device support for TPM-backed full-disk encryption. Authentication integrations with Microsoft Entra ID were strengthened as well.

The release also debuts Myna, Ubuntu's speech-to-text agent, described as an opt-in feature built on local models that provides speech recognition with clear visual feedback while preserving privacy. Alongside that, Canonical expanded architecture support: a dedicated installer for amd64v3 hardware, desktop images for RISC-V users (including support for the SpacemiT K3 RVA23-enabled hardware), and upgraded ROCm to version 10.0 for accelerated AI workloads with support for new GPU architectures and frameworks like vLLM and SGLang.