Ubuntu 26.10 vs 26.04 LTS: Which One Should You Run in 2026?
Two Ubuntu releases matter for almost everyone right now: the just-shipped Ubuntu
26.04 LTS “Resolute Raccoon” (April 2026, 5 years of standard support), and the
upcoming Ubuntu 26.10 “Stonking Stingray” (15 October 2026, 9 months of support).

Both are real, both are good — but they’re for different people. Here’s the
side-by-side.
TL;DR
| If you are… | Run |
|---|---|
| A production server, a stable dev box, a family desktop | 26.04 LTS — stay or install fresh |
| A tester, hobbyist, GNOME 51 enthusiast, RISC-V tinkerer | 26.10 in a VM or as a secondary OS |
| On 24.04 LTS, planning to skip 26.04 entirely | Don’t. LTS-to-LTS is the supported path; jump to 26.04 LTS |
| On 25.10 (already EoL since July 2026) | You’re late — go to 26.04 LTS now, or 26.10 daily if you want cutting edge |
| Buying a new laptop today | 26.04 LTS is the right call |
Support window: the core difference
| 26.04 LTS Resolute Raccoon | 26.10 Stonking Stingray | |
|---|---|---|
| Release date | 2026-04-23 | 2026-10-15 |
| Codename adjective | Resolute | Stonking |
| Support type | LTS (Long-Term Support) | Interim |
| Standard support until | April 2031 (5 years) | July 2027 (9 months) |
| Ubuntu Pro extended | until 2041 (up to 15 years total) | not applicable |
| Recommended upgrade path | 24.04 LTS → 26.04 LTS → 28.04 LTS | 26.10 → 27.04 → 27.10 → 28.04 LTS |
If your machine needs to be solid until 2031 or beyond, 26.04 LTS is the only
choice. 26.10’s support expires before you can upgrade from it to the next LTS
without leaving a gap (you’d have to upgrade 26.10 → 27.04 → 27.10 → 28.04 LTS, and
each interim hop is a 9-month support step that ends before the next LTS ships).
Kernel: 7.0 vs 7.3
26.04 LTS shipped with Linux 7.0 (released April 2026). 26.10 is targeting
Linux 7.3 (still at rc1 at time of writing, but will be near-final by October).
Same kernel lineage, three minor versions apart.

What’s in 7.0 that 26.04 LTS users already have:
- Intel TSX defaults to “auto” mode on capable CPUs (out-of-the-box performance gain)
- AMD EPYC scheduler scalability improvements
- XFS self-healing capabilities
- EXT4 concurrent direct I/O write improvements
- UDP network performance boost from function inlining
- Continued Qualcomm Snapdragon X Elite laptop support
- Intel Crescent Island + initial Nova Lake enablement

What’s new in 7.3 that 26.10 will get:
- 41 million lines of code in the tree (churn from AI/LLM-assisted patch
generation) - Steam Controller (2026 refresh) initial driver support
- AMD Zen 6 Ryzen client enablement
- Intel Xe3P Nova Lake integrated graphics stable + default
- KSMBD Apple Time Machine backup interoperability
- Lenovo ThinkPad USB-C Security ACPI driver support
- Ryzen AI Halo LED/RGB driver
- Scheduler work improving gaming performance on hardware with limited vRAM
- Faster Btrfs across the board (Phoronix benchmarks)
For most desktop users, the practical difference is better hardware support for
brand-new hardware (especially laptops shipping in late 2026 and 2027). If your
machine is from 2024 or earlier, you’ll see essentially no difference between 7.0
and 7.3. If you bought a new Zen 6 / Nova Lake / Snapdragon X2 system this quarter,
you want 7.3.
26.04 LTS users will get 7.3 anyway through the HWE stack in early 2027 with
the 26.04.2 point release. There’s no permanent feature gap.
Desktop: GNOME 50 vs GNOME 51
26.04 LTS shipped with GNOME 50 (Wayland-default, dbus-broker default, AppArmor
profiles tightened). 26.10 ships GNOME 51.
GNOME 51 is a refinement, not a redesign:
| Feature | 26.04 LTS (GNOME 50) | 26.10 (GNOME 51) |
|---|---|---|
| Default desktop | GNOME 50 | GNOME 51 |
| Nautilus “New Document” GUI | not present | added |
| Sushi file previewer | GTK3 | GTK4 + LibAdwaita |
reduced-motion setting |
not present | added |
| Background blur via Wayland | not standardised | ext-background-effect-v1 |
| Fingerprint management UI | basic | new panel |
| Cursor rendering | raster | SVG (HiDPI-clean) |
| systemd-homed initial-setup support | none | initial |
GNOME 51 also removes the legacy NVIDIA proprietary driver code path. If you’re on
a machine with an old nvidia-legacy card, stay on 26.04 LTS — moving to 26.10
without first verifying nvidia-driver-580 works for your card can break graphics.
Authentication: authd + TPM
Both releases share Canonical’s new authd daemon (replacing the old
gnome-remote-desktop authentication flow), and both support TPM-backed full-disk
encryption. 26.04 LTS made it default; 26.10 continues that.
What 26.10 adds: Microsoft password and MFA authentication — log in to your
desktop session with a Microsoft account, with approval requests going through
Microsoft Authenticator. Useful if you live in the Microsoft 365 ecosystem and
don’t want a separate Linux password. (discourse.ubuntu.com)
Architecture support
| 26.04 LTS | 26.10 | |
|---|---|---|
| x86_64 (AMD64) | yes | yes |
| ARM64 (AArch64) | yes | yes |
| RISC-V (RVA23) | server only | full desktop |
If you’re evaluating RISC-V hardware, 26.10 is where you get a usable desktop.
For everyone else, this is a footnote.
App Center: package-agnostic
26.04 LTS introduced the new App Center front-end (replacing the old Ubuntu
Software). 26.10 continues the package-agnostic rewrite — same front-end handles
.deb, snap, and (eventually) Flatpak without treating any one format as special.
For end users this means fewer “this app is a snap” surprises in the UI.
The AI bet: 26.10 starts laying the foundation
This is where 26.10 starts pointing beyond itself. The Context-Aware Desktop
work in 26.10 is the first concrete piece of an architecture that targets
28.04 LTS, not 26.10 itself. The visible deliverable in October: an
on-device speech-to-text engine for the input method framework (offline
voice typing). Everything else — the assistant layer, the app/service/context
fusion — takes more cycles.
26.04 LTS has none of this. It’s a conservative LTS by design: don’t break what
works, ship the platform work, leave the bleeding-edge AI experimentation to
interims.
If “AI features in Ubuntu” is your primary interest, run 26.10 daily. If “stable
production Linux for 5 years” is your primary interest, run 26.04 LTS and ignore
the interim noise.
Boot experience / visual polish
26.04 LTS shipped with a new Plymouth spinner styled after the raccoon tail. 26.10
ships one styled after the stingray tail — same idea, different animal. Wallpapers
follow the same gradient + mascot pattern.
Performance
For most workloads (browsing, office, dev), 26.04 LTS and 26.10 will perform
identically. The 7.3 kernel’s Btrfs and scheduler improvements matter for
specific workloads (gaming on small-vRAM systems, Btrfs-heavy file servers),
not the average desktop.
If you have an NVIDIA card, test on your hardware before committing. NVIDIA
- Wayland + new kernel + new GNOME has historically been the place regressions
appear.
Decision tree
Are you running 24.04 LTS, 22.04 LTS, or a stable production box?
YES → 26.04 LTS, plan to skip 26.10, wait for 28.04 LTS
NO ↓
Are you on 25.10 (which has been EoL since July 2026)?
YES → Upgrade NOW. To 26.04 LTS if you want support, or to 26.10 daily if
you don't mind living on the bleeding edge.
NO ↓
Do you specifically want GNOME 51, Linux 7.3, RISC-V desktop, or Context-Aware
Desktop early?
YES → 26.10 daily in a VM, or dual-boot on a spare partition
NO ↓
You probably want 26.04 LTS.The verdict
26.04 LTS is the right choice for ~95% of users — anyone running a production
server, a stable dev box, a family desktop, or a laptop they want to forget
about until 2031.
26.10 is the right choice for testers and tinkerers — anyone who specifically
wants GNOME 51, the 7.3 kernel’s new hardware support, RISC-V, or wants to be
on the architectural path the Context-Aware Desktop is laying for 28.04 LTS.
The two aren’t competitors. They’re sequential: 26.04 LTS is the bedrock, 26.10
is where the next bedrock’s features get stress-tested. If you’re on 26.04 LTS
already, stay there. If you have a spare partition or a VM, 26.10 is worth a
look.
For the install + upgrade workflow, see the companion piece.
Comments