NEW OpenClaw Update is MASSIVE! (3.12 FREE Release)

OpenClaw 3.12 ships Dashboard v2, Fast Mode, and a wave of security fixes — update today.

Dashboard V2 is the headline

The biggest change in the 3.12 update (released March 12) is a new Dashboard V2 built into the local OpenClaw interface. It gives you a more modular view of agents and sub-agents, ships a new color palette, and adds inline command support. The new layout makes multi-agent workflows easier to scan at a glance.

Fast Mode = surge pricing, not a smarter model

Fast Mode is a /fast toggle that tells OpenAI and Anthropic to prioritize your request. It works for both providers right now, and crucially "fast mode doesn't imply that it's actually dumber" — it's the same model, just queue-jumped. The catch is cost: providers may charge 2x to 4x the standard price for the speed boost. Turn it on when a presentation is due, leave it off the rest of the time.

Local models are now first-class

Ollama support got another round of upgrades, building on the 3.11 onboarding push. OpenClaw, Ollama, and LM Studio all integrate more smoothly now, and providers are now plugin-based, so you can build your own local provider integration via the plugin system. The pitch is straightforward: run local to keep sensitive data off Anthropic's servers, and stop paying per token.

Security is the real reason to update now

OpenClaw took heat in China after a government information department flagged it as insecure (Boxmining has a separate video on the China ban coming). 3.12 ships aggressive patches: device pairing now uses short-lived tokens, workspace plugins can no longer auto-run without permission, and there are fixes for command injection, Unicode tricks, and sandbox escapes. "Your open claw instance is now significantly harder to compromise."

Smaller items worth knowing

Kimmy model performance is noticeably snappier, sub-agent timeouts are now configurable, and Kubernetes support landed for large agent fleets. None of these are reasons to update on their own — the security fixes are.