OpenClaw has released a new version, adding inference, media generation, and other features

BlockBeats News, April 8, OpenClaw released version 2026.4.7, with the following key updates:

· OpenClaw Inference Functionality: Serving as a unified inference workflow hub, it supports provider-driven inference capabilities across models, media, networks, and embedded tasks.

· Tool/Media Generation: Default support for cross-authentication providers (images, music, videos) automatic failover, preserving the original generation intent upon provider switch; automatically mapping parameters such as size, aspect ratio, resolution, duration, etc., to the closest supported options; exposing provider capability query interface; additionally, introducing mode-aware video-to-video transformation support.
· Memory/Wiki: Reinstatement of the built-in memory-wiki technology stack, including plugins, command-line tools, sync/query/apply toolchains, memory host integration, structured statements with evidence fields, compilation digest retrieval, etc.
· Plugin/Webhook: Introduction of built-in Webhook inbound plugin, supporting external automation systems to create and drive binding workflows through independently shared secret endpoints via various routes.

· Support for Arcee, Gemma 4, and Ollama visual models.