v0.6.1
Models Chat Specialists Settings Performance

Cloudflare Workers AI, faster workspaces, fixed mentions

Cloudflare Workers AI joins the model catalog, workspaces no longer block on catalog preload, and explicit specialist mentions go through the join flow again.

Cloudflare Workers AI is now a first-class part of the model catalog, workspaces load even when the catalog is slow or offline, and explicit @mentions once again ask before adding a non-member specialist to your channel.

Cloudflare Workers AI in the catalog

Cloudflare Workers AI’s full catalog is now discoverable alongside OpenRouter, Anthropic, and Bedrock. Workers AI models show up with the right token and context-window details, and Cloudflare-style model names render with cleaner display labels.

If you connect a Cloudflare Workers AI provider in Settings, validation now succeeds on desktop and mobile. The previous “Unsupported provider” error is gone.

Specialist mentions respect channel membership

When you @mention a specialist who is not yet in the channel or thread, you get the add-to-channel confirmation again before they respond. Nothing dispatches until you explicitly approve the join, and a specialist replying no longer silently adds them to the channel. Membership stays explicit, the way the rest of the product expects.

Workspaces stop waiting on the model catalog

Opening a workspace no longer blocks on the model catalog. The workspace UI renders right away and the catalog updates in the background as soon as it’s ready. A slow or failing catalog won’t stall navigation, cached catalogs still refresh, and transient errors stay out of your way.

Polish & fixes

  • Inline tool timeline rows in chat are fully clickable: the chevron and the summary text both toggle the same details panel.
  • GitHub-style emoji shortcodes like :tada: now render in chat, alongside the rest of GitHub-flavored markdown.