The short version: Ubuntu Summit 26.10 opened today, Thursday 12 November 2026, with a free livestream from Canonical’s London base. You can watch it below. Canonical billed Paul Geoghegan’s talk on local AI for Ubuntu accessibility as its big news for Ubuntu, and day one also covered Linux on Snapdragon X2 laptops, attestable VMs and open source AI tooling.
Watch the Ubuntu Summit 26.10 livestream
The official stream: Canonical is broadcasting day one on its Canonical Ubuntu YouTube channel as “Ubuntu Summit 26.10 | Day One”. The stream went live at 09:15 UK time (11:15 SAST), ahead of the 09:30 opening remarks, and the @ubuntu and @Canonical accounts on X have been posting each talk as it starts.
If the player does not load, open the Day One livestream on YouTube directly. To ask speakers questions, sign in to the Ubuntu Summit category on Discourse, where every session has its own thread. Our Ubuntu Summit 26.10 how to watch guide has the full time zone table and watch party details.
The big news: local AI for Ubuntu accessibility
Canonical’s headline talk: ahead of the event, Canonical’s Graham Morrison described “How Local AI is Transforming Ubuntu Accessibility” as “our big news for Ubuntu”. Paul Geoghegan, the Accessibility Engineer for Ubuntu, gave the talk live at 11:30 UK time (13:30 SAST). Geoghegan is a blind engineer, and the session abstract says he works on system-level accessibility and privacy-preserving local AI.
What it covered: the talk argued that cloud models bring privacy risks and fail offline, so assistive tools on a Linux desktop need local, on-device AI. It walked through Project Myna, the team effort to bring native, offline speech-to-text dictation to Ubuntu, and took a broader look at where local AI can practically improve accessibility across the desktop. Myna is already out as a technology preview, and our Myna speech-to-text guide shows how to try it.
From the Q&A: in the talk’s Discourse thread, a blind developer from Egypt asked about Arabic dictation and whether Myna will offer APIs for other apps. Another attendee replied with a screenshot of the Myna model chooser, showing that the Whisper model can be installed and lists Arabic among its languages. A separate question referred to Piper, a text-to-speech engine mentioned in the talk. Canonical has not yet published a written summary, a hardware list or a release date for any new accessibility features, so treat the recording as the primary source for now.
Day one talks at a glance
The full Thursday schedule: these are the sessions from the official timetable. Times are UK time (GMT) and South Africa Standard Time. Recorded talks filled the lunch slot and a short afternoon break.
| UK time | SAST | Session | Speaker(s) |
|---|---|---|---|
| 09:30 | 11:30 | Opening remarks | Jon Seager, Pierre Guillemin |
| 10:15 | 12:15 | Astra: open-source observatory control for exoplanet hunting | Peter Pihlmann Pedersen |
| 10:40 | 12:40 | ntfy: handling 4 million push notifications per day | Philipp Heckel |
| 11:05 | 13:05 | hax: AI coding agent as a lightweight Unix tool | Oleksandr Chekhovskyi |
| 11:30 | 13:30 | How Local AI is Transforming Ubuntu Accessibility | Paul Geoghegan |
| 11:55 | 13:55 | Text to CAD: designing a robot with LLMs | Jake Fitzgerald |
| 14:00 | 16:00 | Pichi: container workflows for portable, attestable VMs | Nathaniel McCallum |
| 14:45 | 16:45 | It Takes an Ecosystem: Building AI in the Open | Katja Sirazitdinova |
| 15:10 | 17:10 | Bringing Delta to the Web with GPUI | Cameron Mcloughlin |
| 15:55 | 17:55 | Linux on Qualcomm Snapdragon X2 platforms | Jason Pettit, Srinivas Kandagatla, Tobias Heider |
| 16:40 | 18:40 | Agentic Computing on the CIX Technology P1 | Jianfeng Liu |
| 17:05 | 19:05 | Lightning talks | Various |
Arm laptops, attestable VMs and open AI
Linux on Snapdragon X2: Jason Pettit and Srinivas Kandagatla of Qualcomm joined Tobias Heider, the Canonical engineer who leads the Ubuntu Concept project, to cover first silicon bring-up, upstream integration, power, graphics and AI work. The abstract promises a “robust, developer-ready Linux experience on modern Arm systems with Ubuntu”, while also naming the challenges that remain.
Agentic computing on the CIX P1: Jianfeng Liu of CIX Technology described the partnership with Canonical to provide native Ubuntu support on the P1 platform, with a Lenovo AI Center Mini case study and future work on the Radxa O6N.
Security and developer tools: AMD’s Nathaniel McCallum presented Pichi, an open source project that gives confidential and conventional VMs a container-style workflow and a Portable Machine Image format. NVIDIA’s Katja Sirazitdinova mapped the open AI stack, from PyTorch and JAX to vLLM, SGLang and Dynamo. Cameron Mcloughlin showed how Zed brought Delta to the browser using GPUI, its GPU-accelerated Rust UI framework.
The morning’s other talks: ntfy founder Philipp Heckel said its free ntfy.sh instance has delivered more than one billion messages in four years and now handles around four million push notifications a day. Oleksandr Chekhovskyi presented hax, an MIT-licensed, terminal-native coding agent written in C that works with local models as well as hosted ones. Jake Fitzgerald showed text-to-cad, his open source library of agent skills, and the six-axis robot arm he designed with it.
What is on Friday and how to keep watching
Day two starts at 09:30 UK time (11:30 SAST) and ends with closing remarks at 17:20 UK time (19:20 SAST). Highlights from the timetable include:
| UK time | SAST | Session | Speaker(s) |
|---|---|---|---|
| 09:30 | 11:30 | EPUBs, ESP32s, and Everyone’s A Developer Now | Justin Mitchell |
| 10:40 | 12:40 | Using Workshop for [agentic] embedded development with Zephyr 26.04 LTS | Viktor Toldov |
| 11:25 | 13:25 | Open source is not enough and why you need attestation | Benny Fuhry, Ijlal Loutfi |
| 15:50 | 17:50 | From Open Weights to Reproducible Results: Inside Nvidia’s Nemotron | Rita Neves |
| 16:15 | 18:15 | Agentic Coding Without the Slop: 700K Lines in 7 Months | Adrian Sieber |
| 17:20 | 19:20 | Closing remarks | Canonical |
Friday also has talks on HTML over WebSockets, personal software, time synchronization and profiling, plus lightning talks. Check the official timetable in case a slot moves. Canonical usually posts Summit talks to YouTube after the event, as it did with our Ubuntu Summit 26.04 recordings. The Summit follows the 26.10 “Stonking Stingray” release, so if you have not upgraded yet, start with what is new in Ubuntu 26.10.
Source: Ubuntu Discourse: Ubuntu Summit Day 1, the wait is almost over!.
