OpenClaw releases updates almost daily. That's not hyperbole. On April 26, another batch landed—and it's a good one. This release tackles three things our users ask about most: making voice conversations feel natural, helping the AI remember what matters, and keeping the assistant reliable across every platform you use. If you're using InstantClaw, you already have all this. If you're self-hosting, it's time to pull the latest.
Generic browser realtime transport contract, Google Live browser Talk sessions with constrained ephemeral tokens, and a Gateway relay for backend-only realtime voice plugins.
Voice conversations with your AI assistant now start faster and stay clearer. The system can use Google's realtime API directly from your browser, without extra relay services. And backend voice plugins get their own dedicated relay so they don't interfere with other tasks.
In human terms: think of the old setup like trying to have a phone conversation through two walkie-talkies pressed together. The new approach is a dedicated conference line where everyone hears clearly, and the system handles the handoffs without dropped words.
If you use voice mode on Telegram, Slack, or a custom app, you'll notice fewer hiccups. The assistant responds faster, interruptions are smoother, and longer conversations don't degrade. It's the difference between talking to someone on a bad cell connection versus a solid landline.
Optional memorySearch.inputType, queryInputType, and documentInputType config for asymmetric embedding endpoints, plus model-specific retrieval query prefixes for Ollama memory-search queries.
The assistant now understands context better when searching its memory. It can use different types of embeddings for questions versus documents (asymmetric), and it knows to treat different memory models differently—like using specialized prefixes for 'nomic-embed-text' vs 'qwen3-embedding'.
In human terms: imagine you have a giant file cabinet. Before, the assistant would just use one search method for everything—like using the same keyword for finding a recipe and a tax form. Now it can use different strategies: for questions, it looks for concepts similar to your query; for documents, it indexes the full text more carefully. It's like having a smart index card system that adapts to what you're looking for.
Your AI assistant will give you better answers when it retrieves past conversations. If you ask 'What did we discuss about the quarterly report?' it's less likely to confuse it with a chat about your vacation. Memory becomes more accurate, especially if you use local models like Ollama.
Bundled Claude and Hermes importers, fixes for Discord thread model overrides, WhatsApp proxy support, Google Meet audio stability, and dozens of other platform-specific crash fixes.
If you're migrating from another AI tool (Claude Code, Hermes), there's now a clean path to bring your setup into OpenClaw. And the platforms you actually use—Discord, WhatsApp, Google Meet, Telegram, Matrix—all get reliability improvements that prevent crashes and miscommunication.
In human terms: it's like finally having a universal remote that actually works with your soundbar, TV, and game console without juggling three remotes. The migration tools are like a moving service that packs your old furniture and sets it up in your new house exactly where you want it.
Switching to OpenClaw from another tool is now realistic—you don't lose your custom instructions, MCP servers, or model settings. And if you rely on your AI assistant in a group chat or a meeting, you'll notice fewer 'oh, it broke' moments. That build trust in using the assistant in real work.
How InstantClaw Users Get Updates Automatically
- Zero effort updates: You never have to download, extract, or reconfigure. Every time OpenClaw releases—sometimes daily—InstantClaw pushes it to your assistant automatically.
- Expert implementation: We handle the tricky parts. Plugin compatibility, migration scripts, config changes—the team behind InstantClaw tests and deploys so you don't have to read release notes unless you want to.
- Continuous improvement: Because updates land frequently, you're always running the most stable, feature-rich version. No waiting for monthly or quarterly releases.
Why Understanding Updates Matters
Even if you don't install updates yourself, knowing what's new helps you understand what your assistant can do now. For example, if you've struggled with voice lag in the past, this update explains why it's suddenly better. Understanding releases lets you spot capabilities you didn't know existed. Maybe you never tried memory search because you thought it was limited—now it's more powerful. It builds informed trust. When you see frequent, substantive improvements, you know the product is alive and invested in.
The Bottom Line
Self-hosting OpenClaw means reading changelogs, pulling Docker images, and hoping nothing breaks. With InstantClaw, you pay for the convenience of always being up to date without the effort. This release alone fixes over 100 issues. That's 100 headaches you didn't have to deal with.
Want OpenClaw updates without the maintenance?
Deploy in under a minute. No SSH. No updates. No 3am patches.
InstantClaw