[Release] openclaw/openclaw v2026.4.15: openclaw 2026.4.15

OpenClaw v2026.4.15 is a major release featuring Claude Opus 4.7 as the default Anthropic model, new Google Gemini text-to-speech support, a Model Auth status card for OAuth health monitoring, cloud storage support for LanceDB memory, and GitHub Copilot embedding provider integration. The release includes numerous stability fixes across gateway tools, agent replay recovery, Docker builds, Matrix E2EE, memory management, and channel integrations, plus packaging improvements to reduce bundle size and plugin runtime dependencies.

Key Points

• •Claude Opus 4.7 now default for Anthropic selections with bundled image understanding capabilities
• •Google Gemini text-to-speech added to bundled plugin with WAV and PCM telephony output support
• •New Model Auth status card displays OAuth token health and provider rate-limit pressure with expiration alerts
• •LanceDB memory now supports cloud storage for durable remote indexes instead of local-disk-only
• •GitHub Copilot embedding provider added for memory search with safer payload validation and token refresh
• •Experimental `agents.defaults.experimental.localModelLean` mode reduces prompt size by dropping heavyweight default tools
• •Plugin runtime dependencies localized to extensions; published builds trimmed and leaner
• •Gateway tool security: trusted MEDIA passthrough anchored to exact registered tool names; client tools rejected if names collide with built-ins
• •Memory management improvements: default dreaming storage changed from inline to separate files; memory excerpts capped with continuation metadata
• •Multiple integration fixes: Matrix E2EE bootstrap, Discord tool-call text stripping, WhatsApp auth creds race condition, BlueBubbles catchup retry ceiling, Ollama model ID prefix handling

Workflow Diagram

Concepts

Artifacts (3)

agents:
  defaults:
    experimental:
      localModelLean: true

plugins:
  entries:
    memory-core:
      config:
        dreaming:
          storage:
            mode: "separate"

catchup:
  maxFailureRetries: 10