update
Apr 28, 2026
By Teun
Ollama v0.22.0 adds Nemotron 3 Omni, Laguna XS.2
Ollama released v0.22.0 on 28 April, adding support for NVIDIA’s Nemotron 3 Omni and Poolside’s Laguna XS.2. The release also includes new installers and platform builds for macOS, Linux, Windows, and JetPack variants, according to the GitHub release page.
Ollama has released version v0.22.0, with the update posted on 28 April by maintainer dhiltgen on GitHub. According to the release page, the new version adds support for two new models: NVIDIA’s Nemotron 3 Omni and Poolside’s Laguna XS.2.
The release is part of the open-source Ollama project, which is widely used to run large language models locally. While the GitHub post is short on feature details, it clearly frames the update around expanded model support rather than a long list of runtime changes or API additions.
⚡ New to this?
This is a new software release for Ollama, a tool that lets people run AI models on their own machines or servers. A model is the AI system itself; adding new models means users have more choices for local automation and testing.
The acronyms and platform names matter because Ollama ships different builds for Linux, Windows, macOS, and NVIDIA GPU setups like ROCm and JetPack. That affects which hardware can run the software without extra setup.
🦞 OpenClaw angle
If you self-host Ollama in automation flows, update your pinned version and test v0.22.0 in a staging environment before rolling it into production. Check whether Nemotron 3 Omni or Laguna XS.2 fits any of your agent tasks, then compare outputs against your current model on the same prompts.
Also verify the new installer or binary matches the checksum file before deployment. If you run mixed hardware, match the package to the target platform, especially for ROCm, JetPack, or ARM builds.
According to the release notes, the update also ships a broad set of binaries and installers for different systems. These include install scripts for PowerShell and shell environments, builds for macOS, Linux, Windows, and multiple ARM and ROCm variants, plus JetPack-specific Linux packages.
The assets listed on GitHub include Ollama installers and archives for Darwin, Windows, and Linux, along with checksum files. GitHub also shows source archives for the v0.22.0 tag, which lets operators verify downloads or build from source if they prefer.
The release page says v0.22.0 was published at 15:00 on 28 April, and the tag points to commit 955112e. GitHub lists five commits to main since that release, which suggests work continued immediately after the tag was cut.
Ollama did not spell out model-specific changes in the release post itself. The announcement simply says “New models” and names the two additions, with links to the Ollama model library pages for Nemotron 3 Omni and Laguna XS.2.
For teams running local model infrastructure, that means the main change in v0.22.0 is access to additional models in the Ollama ecosystem. The release page does not mention deprecations, breaking changes, or security fixes, so the announcement is narrowly focused on model availability and packaged builds.
The full changelog link on the release page points back to the comparison between v0.21.2 and v0.22.0. GitHub also displays the usual release assets and SHA-256 checksums, which can be used to confirm that downloaded files match the published release.
In short, Ollama v0.22.0 is now available with support for Nemotron 3 Omni and Laguna XS.2, plus updated installers and build artifacts across supported platforms, according to the GitHub release.
