—

 

 

 

OpenClaw 2026.4.29 🦞

 

💬 Group chats feel much better now

📌 Follow-up commitments from context

🔐 Safer exec, pairing, and owner controls

🟩 NVIDIA provider + model catalogs

⚡ Faster startup + plugin/channel fixes

 

Group chat finally feels agent-native.

https://github.com/openclaw/openclaw/releases/tag/v2026.4.29

 

 

Follow-up commitments are opt-in: OpenClaw can infer lightweight “check back on this later” items from conversation context, then heartbeat delivers them when due. https://docs.openclaw.ai/concepts/commitments

 

 

Group chats now feel more natural: the agent can think, use tools, then deliberately send or edit the visible message instead of having the chat loop post the final assistant text. https://docs.openclaw.ai/channels/groups

 

 

Model setup got smoother too: NVIDIA joins as a first-class provider with catalog-backed picks for Nemotron, Kimi, MiniMax, GLM, and friends. https://docs.openclaw.ai/providers/nvidia

 

 

Exec and pairing controls got less squishy: restrictive profiles stay restrictive, owner checks are tighter, and startup warnings call out risky configs early. https://docs.openclaw.ai/tools/exec-approvals

 

 

Memory got more useful in real conversations: people/wiki metadata, provenance, aliases, relationship graphs, scoped recall, and partial results when recall times out. https://docs.openclaw.ai/plugins/memory-wiki

 

 

Startup and plugin installs should feel less haunted: better slow-host diagnostics, event-loop readiness signals, runtime-dependency repair, and a broad Slack/Telegram/Discord/WhatsApp reliability pass. https://docs.openclaw.ai/cli/gateway

 

 

Busy chats got calmer: follow-ups can steer the active run at the next model boundary instead of spawning duplicate work or waiting for the loop to finish first. https://docs.openclaw.ai/concepts/queue

 

 

@SolomonWis54084 We fixed a lot around plugins, yes.

 

 

@rushdee1776 agents: {

defaults: {

experimental: {

localModelLean: true

}

}