OpenClaw
Dynamic model discovery lets OpenClaw add newly launched AI models without waiting for an OpenClaw update. The model picker now draws from a hosted catalogue, the user’s provider account and their installed Codex.
The change follows a launch-day problem with GPT-6.1 Sol. People using OpenClaw with a ChatGPT sign-in could not select the model because OpenClaw bundled Codex 0.158.0, while Sol required Codex 0.159.2 or newer. Changing that version previously required a full OpenClaw release.
OpenClaw rebuilt its model discovery system over about five weeks and roughly 80 pull requests, according to the OpenClaw blog post.
Dynamic model discovery uses three sources
Previously, each provider plugin carried a model list in a manifest file inside the OpenClaw package. Because plugins were tied to releases, adding a model meant editing the manifest, publishing a release and asking users to update.
| Source | What it contributes | How it updates |
|---|---|---|
| Hosted catalogue | Model data, OpenClaw manifests and prices | Rebuilt every four hours |
| Provider account | The account’s available model list | Read from the provider |
| Installed Codex | Models supported by the installed Codex version | Read from the local installation |
It starts with data from models.dev, overlays OpenClaw’s own manifests, adds pricing and publishes the result at catalog.openclaw.ai.
A running OpenClaw Gateway checks for newer catalogue data in the background. When it finds an update, it switches to the new catalogue while running. A Gateway restart is no longer required to pick up that catalogue change.
Model launches no longer wait for OpenClaw releases
The new route removes OpenClaw’s release cycle from the model launch sequence. Once a model appears through one of the three discovery sources, it can reach the model picker without being packaged into a new OpenClaw version.
This changes the old sequence from provider launch, manifest edit, OpenClaw release and user update to direct discovery from the catalogue, provider account or installed Codex. It also separates updates to model availability from updates to the OpenClaw application itself.
Frequently asked questions
What is Dynamic model discovery in OpenClaw?
Dynamic model discovery is the system OpenClaw uses to find available AI models without requiring a new application release. It combines data from a hosted catalogue, the user’s provider account and their installed Codex.
When is Dynamic model discovery available?
The blog post does not give a release date. Running Gateways check for newer hosted catalogue data in the background and can switch to it without restarting.
Which OpenClaw plan includes Dynamic model discovery?
The blog post does not list plan or pricing restrictions. It describes the feature as part of OpenClaw’s model discovery system.
Sources
3 checkedHow we cover tool news: Create With's tool desk drafts these reports with AI from the sources listed above and checks them against those sources before publishing.
