This is a pretty big one. Not just small tweaks… there are some meaningful upgrades that change how things work day-to-day. Here are the highlights worth knowing: ⚠️ Breaking changes (important) - Qwen OAuth via portal.qwen.ai has been removed - You now need to use Model Studio with API key auth - Older configs (over 2 months) won’t auto-fix anymore — they’ll fail validation instead If you’ve got older setups running, it’s worth double-checking things. 🔧 New features & improvements - xAI integration upgraded (better search + smoother setup) - Image generation added via MiniMax (including image-to-image editing) - Plugins can now request approval before running tools (huge for control) - apply_patch now enabled by default for OpenAI/Codex models - Better onboarding flows for web search + tools 💡 Platform improvements - Cleaner plugin system (less manual setup needed) - Better CLI backend support (including Gemini CLI) - File uploads and messaging actions becoming more unified across platforms - Improvements to container setup and workflows 🛠 Fixes (lots of them)There’s a long list, but the big wins are: - More stable agent runs (fewer crashes) - Better handling of API errors and rate limits - Fixes for Telegram, WhatsApp, Discord, and more - Improved image handling across providers - Cleaner UI behaviour in multiple areas My takeThis update is all about stability + control. Less “it kinda works”More “it works properly and predictably” Especially with: - plugin approvals - better error handling - improved integrations If you’re building anything serious with OpenClaw, this is a solid step forward. If you’ve updated already, I’d be interested to hear what you’ve noticed. Cheers Jason

If you’re actively using OpenClaw, this latest release is actually a meaningful step forward… not just small tweaks. A few things stood out to me 👇 1. Better OpenAI / RAG compatibility They’ve added /v1/models and /v1/embeddings, plus improved how model overrides work. 👉 In plain terms: If you're building anything with RAG, embeddings, or external tools… things should just connect more smoothly now. 2. Tools are finally clearer (big one) The /tools endpoint now shows what your agent can actually use right now. Plus there’s a new “Available Right Now” section in the UI. 👉 This removes a lot of guesswork.Before, you’d try something and hope it worked… now you can see it upfront. 3. Teams + Slack got a serious upgrade - Microsoft Teams now uses the official SDK with proper AI UX - Slack replies are cleaner and interactive elements work better 👉 If you're building bots for clients, this makes things feel much more “production ready”. 4. Skills are much easier to install They’ve added one-click install recipes for things like: - coding agent - GitHub issues - Whisper - Trello - weather 👉 This is a big step toward making OpenClaw usable for non-devs (or at least less painful). 5. Docker + CLI improvements You can now run OpenClaw commands directly inside containers with --container. 👉 Useful if you're hosting on a VPS or running multiple environments. 6. Loads of real-world fixes There are a ton of fixes across: - Telegram - WhatsApp - Discord - Slack - Docker setups 👉 Basically… fewer random breakages and more reliability. My takeaway This update feels like OpenClaw is moving from “powerful but a bit rough around the edges”to “something you could actually rely on for real deployments”. The direction is clear now: - better integrations - better UX - less friction setting things up If you're using it already it's worth pulling the latest version. Cheers Jason 🙌

OpenClaw v2026.3.8 is out 🦞 Another quick iteration from the OpenClaw team and this one continues the same theme we’ve been seeing throughout the 2026 cycle… Less “look at this shiny feature”More tightening of the core engine. After digging through the changes, here’s what actually matters in practical terms. ⚡ Reliability improvements across the agent runtime A lot of work in this release focuses on making the runtime more predictable. That means fewer weird states where agents stall, hang, or behave inconsistently during long sessions. If you’re running OpenClaw as a daily driver (especially on a VPS), stability improvements like this tend to be the most noticeable over time. 🔁 Messaging and delivery handling More refinements around message delivery and queue handling. In normal terms: • fewer stuck messages • cleaner retry behaviour • better handling when integrations briefly fail If you’ve ever had a bot message silently fail or get trapped in a loop, updates like this are aimed directly at that. 🧠 Agent lifecycle and context improvements There are more tweaks around how agents spin up, handle context, and terminate sessions. Translation: • cleaner startup behaviour • less leftover state from previous runs • more predictable responses during longer tasks These small lifecycle fixes are what gradually turn experimental agents into something you can actually rely on. 🔐 Continued security hardening Security tightening continues across the workspace and execution environment. The general direction is clear:OpenClaw is being hardened so agents can run with real permissions without creating risky edge cases. If you’re experimenting with automation, that’s an important trend. 📱 Ecosystem support keeps expanding The broader messaging and device ecosystem is still being refined. That includes ongoing work around integrations and node/device capabilities. For anyone running OpenClaw across Telegram, Discord, or multi-agent setups, these improvements keep making the system smoother.

A new OpenClaw release landed with improvements across the dashboard, models, plugins, agents, and security. New Control Dashboard 🧭 - Refreshed modular layout and clearer views for overview, chat, config, agents, sessions - Command palette, chat search/export, pinning and mobile-friendly navigation Fast Mode for OpenAI + Claude ⚡ - Fast Mode toggle across /fast, UI, TUI and ACP - Correct mapping to provider fast tiers and request shaping Provider Plugin Architecture 🔌 - Ollama, vLLM and SGLang now use provider plugin onboarding and model discovery Kubernetes Deployment Docs ☸️ - Starter raw manifests and docs for easier scaling Agent Improvements 🧠 - New sessions_yield capability to end the turn immediately while carrying payload forward Slack Improvements 💬 - Agents can send Slack Block Kit messages through standard reply system Security Improvements 🔒 - Short-lived bootstrap tokens, safer plugin loading defaults, improved approvals and scope protection Quick takeaway ✅ Better dashboard experience, faster execution, stronger architecture + security. Cheers Jason

OpenClaw v2026.3.2 is out, and after going through the release notes, this one continues the 2026 trend: less flashy features, more tightening of the core engine. Here’s what stands out in practical terms: 🔐 Secrets workflow improvements The newer structured secrets management system continues to mature. That means: - Cleaner API key handling - Safer apply/reload behavior - Less risk of config drift or messy upgrades If you’ve ever had weird auth issues after an update, this direction is a good sign. 🧵 Thread-bound / ACP agent stability Thread sessions and agent lifecycle handling keep getting refined. In normal language: - Fewer stuck thread states - Cleaner agent startup and cleanup - More predictable replies inside threaded environments If you’re running Telegram, Slack, or multi-agent setups, this matters. 🔁 Delivery queue & retry reliability There are more fixes around retry backoff and queue handling. Translation:Failed sends are less likely to create weird starvation loops or get stuck in awkward states. That’s the kind of fix you don’t notice… until you don’t have random delivery glitches anymore. ⌨️ Typing indicator fixes (again, but deeper) They’ve continued hardening typing cleanup logic across channels. This includes: - Preventing stuck “typing…” states - Cross-channel typing isolation - Time-to-live safety guards If you’ve ever had a bot sit there “typing” forever, this is directly aimed at that. 📱 Device & node support continues to expand Android node commands and device capability handling are improving.The broader 2026.x cycle has also been strengthening mobile/node onboarding. If you’re building a proper node mesh setup, this release keeps pushing that forward. 🛡️ Safety & enforcement handling (Telegram especially) There are bounded backoffs and protections to avoid aggressive retry loops that could trigger platform enforcement. That’s important if you’re running production bots. Big Picture v2026.3.2 feels like another “make the engine solid” release.