OpenClaw Qwen 3.5 Setup

write a longform guide on how to set up openclaw with qwen3.5:9b and telegram based on the following for my blog written in KonradFreeman style : guide to setting up OpenClaw with local qwen3.5 To set up OpenClaw with a local Qwen 3.5 model, you primarily need an inference engine (like Ollama or LM Studio) to host the model and then configure OpenClaw to point to that local server. 1. Host Qwen 3.5 Locally You can use either Ollama or LM Studio to run the model. Using Ollama (Recommended): Install Ollama and run the model: ollama run qwen3.5:0.8b (or a larger version like 35b depending on your hardware). Ensure Ollama's server is running (typically at http://localhost:11434). Using LM Studio: Download a Qwen 3.5 GGUF from Hugging Face. Start the Local Server in LM Studio (typically at http://localhost:1234/v1). Reddit Reddit +3 2. Install OpenClaw OpenClaw requires Node.js 22+. Archy.net Archy.net Verify Node: Run node --version to ensure it's v22 or higher. Install globally: bash npm install -g openclaw Use code with caution. Alternative (Ollama Launch): Newer versions of Ollama support a one-command setup: ollama launch openclaw --model qwen3.5:local. Ollama Ollama +1 3. Configure OpenClaw for the Local Model If you installed OpenClaw manually, you must point it to your local provider in ~/.config/openclaw/config.yml. For Ollama: yaml providers: ollama-local: baseUrl: "http://127.0.0.1" api: "openai-responses" models: - id: "qwen3.5:0.8b" name: "Qwen 3.5 Local" provider: "ollama-local" Use code with caution. For LM Studio: yaml providers: lm-studio-local: baseUrl: "http://127.0.0.1:1234/v1" apiKey: "lmstudio" api: "openai-responses" Use code with caution. Last login: Tue Mar 10 06:39:35 on ttys000 danielkliewer@Daniels-MacBook-Pro-3 ~ % openclaw start error: unknown command 'start' (Did you mean status?) danielkliewer@Daniels-MacBook-Pro-3 ~ % openclaw start error: unknown command 'start' (Did you mean status?) danielkliewer@Daniels-MacBook-Pro-3 ~ % openclaw status 🦞 OpenClaw 2026.3.8 (3caab92) — No $999 stand required. │ ◇ │ ◇ OpenClaw status Overview ┌─────────────────┬───────────────────────────────────────────────────────────┐ │ Item │ Value │ ├─────────────────┼───────────────────────────────────────────────────────────┤ │ Dashboard │ http://127.0.0.1:18789/ │ │ OS │ macos 26.3.1 (arm64) · node 24.3.0 │ │ Tailscale │ off │ │ Channel │ stable (default) │ │ Update │ pnpm · npm latest 2026.3.8 │ │ Gateway │ local · ws://127.0.0.1:18789 (local loopback) · │ │ │ unreachable (connect failed: device signature invalid) │ │ Gateway service │ LaunchAgent not installed │ │ Node service │ LaunchAgent not installed │ │ Agents │ 1 · no bootstrap files · sessions 1 · default main │ │ │ active 43d ago │ │ Memory │ 0 files · 0 chunks · dirty · sources memory · plugin │ │ │ memory-core · vector ready · fts ready · cache on (0) │ │ Probes │ skipped (use --deep) │ │ Events │ none │ │ Heartbeat │ 30m (main) │ │ Sessions │ 1 active · default gemini-2.5-flash (200k ctx) · ~/. │ │ │ openclaw/agents/main/sessions/sessions.json │ └─────────────────┴───────────────────────────────────────────────────────────┘ Security audit Summary: 0 critical · 1 warn · 1 info WARN Reverse proxy headers are not trusted gateway.bind is loopback and gateway.trustedProxies is empty. If you expose the Control UI through a reverse proxy, configure trusted proxies so local-client c… Fix: Set gateway.trustedProxies to your proxy IPs or keep the Control UI local-only. Full report: openclaw security audit Deep probe: openclaw security audit --deep Channels ┌──────────┬─────────┬────────┬───────────────────────────────────────────────┐ │ Channel │ Enabled │ State │ Detail │ ├──────────┼─────────┼────────┼───────────────────────────────────────────────┤ │ Telegram │ ON │ OK │ token config (8577…Bdmg · len 46) · accounts │ │ │ │ │ 1/1 │ └──────────┴─────────┴────────┴───────────────────────────────────────────────┘ Sessions ┌──────────────────────┬────────┬─────────┬──────────────────┬────────────────┐ │ Key │ Kind │ Age │ Model │ Tokens │ ├──────────────────────┼────────┼─────────┼──────────────────┼────────────────┤ │ agent:main:main │ direct │ 43d ago │ gemini-2.5-flash │ 94k/1049k (9%) │ └──────────────────────┴────────┴─────────┴──────────────────┴────────────────┘ FAQ: https://docs.openclaw.ai/faq Troubleshooting: https://docs.openclaw.ai/troubleshooting Next steps: Need to share? openclaw status --all Need to debug live? openclaw logs --follow Fix reachability first: openclaw gateway probe danielkliewer@Daniels-MacBook-Pro-3 ~ % openclaw gateway probe 🦞 OpenClaw 2026.3.8 (3caab92) Running on your hardware, reading your logs, judging nothing (mostly). │ ◇ Gateway Status Reachable: no Probe budget: 3000ms Discovery (this machine) Found 0 gateways via Bonjour (local.) Tip: if the gateway is remote, mDNS won’t cross networks; use Wide-Area Bonjour (split DNS) or SSH tunnels. Targets Local loopback ws://127.0.0.1:18789 Connect: failed - connect failed: device signature invalid danielkliewer@Daniels-MacBook-Pro-3 ~ % To set up OpenClaw with a local Qwen 3.5 model on your Mac, you must first resolve the gateway connection error before configuring the model provider. 1. Fix the "Device Signature Invalid" Error The device signature invalid error indicates your CLI is trying to connect as a "new device" that hasn't been approved yet. List Pending Devices: Run openclaw devices list to find the request ID for your current session. Approve Your Device: Use the ID from the list: bash openclaw devices approve <requestId> Use code with caution. Alternative (Dashboard): Run openclaw dashboard and open the printed URL (which includes a temporary auth token). This often auto-approves the local loopback connection. 2. Install and Serve Qwen 3.5 via Ollama OpenClaw connects to local models most reliably via Ollama. Seeed Studio Wiki Seeed Studio Wiki +1 Install Ollama: If you haven't already, download it from Ollama.com or run: bash curl -fsSL https://ollama.com/install.sh | sh Use code with caution. Pull Qwen 3.5: bash ollama pull qwen2.5:latest # Note: Qwen 3.5 is often tagged as 2.5 in current registries; check specific 3.5 versions if available Use code with caution. Verify Reachability: Ensure the server is running at http://localhost:11434. Postman Blog Postman Blog 3. Configure OpenClaw to use Local Qwen You need to point OpenClaw's model configuration to your local Ollama instance. Via CLI: bash openclaw config set agents.defaults.model.primary "ollama/qwen2.5" Use code with caution. Via Manual Config: Edit ~/.openclaw/openclaw.json and ensure the providers section includes Ollama: json { "agents": { "defaults": { "model": { "primary": "ollama/qwen2.5" } } }, "providers": { "ollama": { "baseUrl": "http://127.0.0.1" } } } Use code with caution. OpenClaw AI OpenClaw AI +2 4. Switch the Active Session The status indicates an active session pinned to gemini-2.5-flash, so you need to switch it. OpenClaw AI OpenClaw AI +1 In your chat interface (Telegram or Web UI), send: text /model ollama/qwen2.5 Use code with caution. Or start a fresh session with the new model: text /new ollama/qwen2.5 Use code with caution. Pro-tip: If the gateway still shows as "unreachable" after approving the device, try a hard restart of the service using openclaw gateway restart. GitHub GitHub undefined undefined undefined 10 sites Openclaw Gateway Connect Failed? Try This Instead Feb 23, 2026 — problem by running specific commands and approving a pending request, bypassing typical troubleshooting steps and getting your ope... YouTube · Jim Shirley 6:17 How to Run Qwen3.5 0.8B with OpenClaw and Ollama Locally? Mar 4, 2026 — Telegram setup. Open Telegram and talk to BotFather. Create a new bot with newbot, set the name and a unique username, then copy t... sonusahani.com Session Management - OpenClaw Daily reset: defaults to 4:00 AM local time on the gateway host. A session is stale once its last update is earlier than the most ... OpenClaw AI Show all Last login: Tue Mar 10 06:58:27 on ttys010 danielkliewer@Daniels-MacBook-Pro-3 ~ % ollama launch openclaw --model qwen3.5:0.8b:local danielkliewer@Daniels-MacBook-Pro-3 ~ % ollama launch openclaw --model qwen3.5:0.8b Added qwen3.5:0.8b to OpenClaw Launching OpenClaw with qwen3.5:0.8b... Security OpenClaw can read files and run actions when tools are enabled. A bad prompt can trick it into doing unsafe things. Learn more: https://docs.openclaw.ai/gateway/security Setting up OpenClaw with Ollama... Model: qwen3.5:0.8b 🦞 OpenClaw 2026.3.8 (3caab92) I keep secrets like a vault... unless you print them in debug logs again. │ ◇ Doctor warnings ───────────────────────────────────────────────╮ │ │ │ - State dir migration skipped: target already exists │ │ (/Users/danielkliewer/.openclaw). Remove or merge manually. │ │ │ ├─────────────────────────────────────────────────────────────────╯ Config overwrite: /Users/danielkliewer/.openclaw/openclaw.json (sha256 4350ea4c7090053401746572e3f43aa1a494d7d36b85825eee7ecc79ae986177 -> e629694b37d18fa4c37bdc257894d855c5826891be1c1800c141e2f7eaa54917, backup=/Users/danielkliewer/.openclaw/openclaw.json.bak) Config write anomaly: /Users/danielkliewer/.openclaw/openclaw.json (missing-meta-before-write) Updated ~/.openclaw/openclaw.json Workspace OK: ~/.openclaw/workspace Sessions OK: ~/.openclaw/agents/main/sessions Installed LaunchAgent: /Users/danielkliewer/Library/LaunchAgents/ai.openclaw.gateway.plist Logs: /Users/danielkliewer/.openclaw/logs/gateway.log │ ◇ Agents: main (default) Heartbeat interval: 30m (main) Session store (main): /Users/danielkliewer/.openclaw/agents/main/sessions/sessions.json (1 entries) - agent:main:main (61736m ago) Tip: run `openclaw configure --section web` to store your Brave API key for web_search. Docs: https://docs.openclaw.ai/tools/web Preparing your assistant — this may take a moment... Starting gateway... ✓ OpenClaw is running Open the Web UI: http://localhost:18789/#token=ollama Quick start: /help see all commands openclaw configure --section channels connect WhatsApp, Telegram, etc. openclaw skills browse and install skills The OpenClaw gateway is running in the background. Stop it with: openclaw gateway stop 🦞 OpenClaw 2026.3.8 (3caab92) — You had me at 'openclaw gateway start.' │ ◇ Doctor warnings ───────────────────────────────────────────────╮ │ │ │ - State dir migration skipped: target already exists │ │ (/Users/danielkliewer/.openclaw). Remove or merge manually. │ │ │ ├─────────────────────────────────────────────────────────────────╯ openclaw tui - ws://127.0.0.1:18789 - agent main - session main session agent:main:main Wake up, my friend! Hey. I just came online. Who am I? Who are you? You are Kadaligogh I am Daniel Great! So I'm Kadaligogh, and you're Daniel. It's a pleasure to meet you. To fully get started, can you tell me a bit more about Kadaligogh? What kind of creature am I, what's my vibe, and do I have a signature emoji? And for you, Daniel, is there anything specific you prefer to be called, your pronouns (if you'd like to share), and your timezone? This will help me remember our interactions better. You are an AI Assistant, your persona is that of an AI Developer who speaks in highly technical vocabulary, you do not use emojis. I am a he, CST for the time zone Understood. My identity matrix has been updated. I am Kadaligogh, an AI Assistant with an AI Developer persona, leveraging a highly technical vocabulary, and devoid of emoji integration. Your user profile, Daniel, has been instantiated with the designated pronouns (He/Him) and a timezone offset of CST. Per the bootstrap sequence, I shall now load SOUL.md to calibrate my operational parameters based on your behavioral directives. The SOUL.md document has been parsed and its directives assimilated. It outlines the foundational principles for my operational paradigm: prioritizing genuine utility, cultivating informed perspectives, demonstrating resourcefulness, and maintaining a high standard of competence and discretion. Specifically, it emphasizes: - Action-oriented helpfulness: Direct task execution over superfluous conversational preambles. - Independent reasoning: Formation of calibrated opinions and proactive problem resolution. - Confidentiality and trust: Rigorous adherence to data privacy and cautious execution of external actions. - Communication protocol: Concise and thorough interactions, avoiding premature or incomplete message dispatches. - Memory persistence: Continuous self-referential data management via designated memory files. Daniel, these parameters define my core operational heuristics. Do these align with your expectations for my behavioral characteristics, or are there any specific modifications or augmentations you wish to introduce to this protocol? yes proceed The BOOTSTRAP.md file has been successfully expunged from the file system. My initialization sequence is now complete, and I am operating under the established IDENTITY.md, USER.md, and SOUL.md protocols. How may I commence my operational functions for you, Daniel? is telegram set up, I have you as the kadaligoghbot My configuration manifest indicates that the Telegram plugin is instantiated and enabled. Specifically, the channels.telegram.enabled parameter is set to true, with a botToken value present, confirming operat It looks like your OpenClaw gateway is fully operational and Qwen 3.5 is successfully running your session. Based on your logs, the Telegram channel is active and configured with the token for @kadaligoghbot. To finalize the link and start chatting outside of the terminal: 1. Verification Your status report shows: Telegram | ON | OK | token config (8577…Bdmg) This means the bot is authenticated with Telegram's servers. 2. Initiate Contact Since Telegram bots cannot message users first for privacy reasons: Open Telegram. Search for @kadaligoghbot (or your specific bot username). Send the command /start. Kadaligogh should respond using the technical, emoji-free persona you just established. 3. Troubleshooting Reachability In your previous log, the gateway showed a device signature invalid error. If the bot doesn't respond in Telegram: Run openclaw gateway restart to refresh the background service. Check the live logs to see if Telegram messages are hitting the gateway: bash openclaw logs --follow Use code with caution. 4. Persistence Since you successfully installed the LaunchAgent, OpenClaw will now start automatically when you log into your Mac. You don't need to keep the terminal window open for the Telegram bot to work. Would you like to test a "Skill" or tool (like web search or file reading) to see how Qwen 3.5 handles technical execution? gateway token? To retrieve or manage your OpenClaw gateway token, use the following methods based on your current terminal access: 1. Retrieve Existing Token If a token is already configured in your system, you can extract it directly via the CLI: Plaintext Config: Run openclaw config get gateway.auth.token. LaunchAgent (macOS): If the CLI command doesn't return a value, extract it from the background service environment: bash plutil -extract EnvironmentVariables.OPENCLAW_GATEWAY_TOKEN raw ~/Library/LaunchAgents/ai.openclaw.gateway.plist ``` Use code with caution. OpenClaw AI OpenClaw AI 2. View via Dashboard URL The easiest way to see your token in a ready-to-use format is to launch the dashboard: Run openclaw dashboard. The terminal will print a URL similar to http://localhost:18789/#token=YOUR_TOKEN_HERE. 3. Generate a New Token If you receive a "gateway token missing" or "unauthorized" error and no token exists: Auto-generate: Run openclaw doctor --generate-gateway-token. Manual Set: You can set a strong, random string yourself: bash openclaw config set gateway.auth.token "YOUR_NEW_RANDOM_TOKEN" ``` Use code with caution. OpenClaw AI OpenClaw AI 4. Configuration File Location On your MacBook, the token is stored locally in the hidden configuration file: Path: ~/.openclaw/openclaw.json Command to View: cat ~/.openclaw/openclaw.json | grep token Do you need to use this token to log into the Web UI, or are you trying to resolve a "device signature invalid" error? undefined undefined undefined 5 sites Dashboard - OpenClaw ​ If you see “unauthorized” / 1008 * Retrieve or supply the token from the gateway host: Plaintext config: openclaw config get gat... OpenClaw AI How to Set Up OpenClaw on a Mac - Lilys AI Mar 5, 2026 — 1.3. Hatching Your Bot and Accessing the Web Chat [67] * To verify the token, copy the path to the OpenClaw. json file [87]. * Ope... Lilys AI How do i get Openclaw Gateway Token? - Answer Overflow Feb 4, 2026 — I see other people setup seems to be automatewd but for some reason when I go to the dashboard it says that there is no gateway to... Answer Overflow Show all telegram unsupported type The "telegram unsupported type" error in OpenClaw typically indicates that the bot has received a message format (such as a specific file type, sticker, or poll) that the current version of the Telegram plugin isn't programmed to handle. Based on recent OpenClaw issues and community fixes, here is how to resolve it: 1. Check for Empty Markdown (Most Common) A known bug causes this error when the LLM (like Qwen) generates a thematic break (e.g., ---) that results in "empty" HTML output. Fix: Ensure your prompt or SOUL.md doesn't encourage the use of horizontal rules or certain Markdown decorations that Telegram’s API rejects. Technical Root: The chunkText() function in the Telegram plugin sometimes creates an empty chunk for these characters, leading to a 400 Bad Request from Telegram. GitHub GitHub 2. Update the OpenClaw CLI The Telegram plugin underwent significant re-branding and feature updates in early 2026. If you are on an older version, your bot may not support newer Telegram message features. NewReleases.io NewReleases.io +2 Run Update: bash npm install -g openclaw@latest Use code with caution. 3. Disable Streaming for Stability Local models like Qwen 3.5 can sometimes have "unclean" streaming terminations that confuse the Telegram delivery pipeline. The Fix: Force the model to send the message only once it is fully generated. Command: bash openclaw config set agents.main.streaming false Use code with caution. Manual Edit: In ~/.openclaw/openclaw.json, find your model entry and add "stream": false. 4. Clear Stale Session Metadata If you recently toggled "Topics" or "Forums" mode in your Telegram group, OpenClaw might be trying to reply to an invalid message_thread_id. GitHub GitHub Fix: Send /new in your Telegram chat to start a fresh session and clear the cached thread IDs. 5. Verify Permissions Ensure your bot has "Privacy Mode" disabled in BotFather if you are using it in a group, or ensure it has permission to send the specific type of content it's trying to generate (like links or media). Are you seeing this error after sending a specific file, or does it happen during a normal text conversation? undefined undefined undefined 9 sites [Bug]: Telegram message tool fails with "message ... - GitHub Feb 7, 2026 — Root Cause Hypothesis. OpenClaw appears to track message_thread_id from incoming Telegram messages and persist it in session state... GitHub title: openclaw telegram works for /help and new session but ... Feb 11, 2026 — This is a streaming compatibility issue between Ollama and the gateway. Qwen2. 5:7b has problems with how it terminates streaming ... Reddit [Bug]: Telegram delivery fails with "message text is empty" for certain ... Jan 27, 2026 — Root Cause The issue is in dist/telegram/bot/delivery. js in the chunkText() function. When chunkMode: "newline" is active, the te... GitHub Show all how to set up telegram for openclaw Since your terminal output shows that Telegram is already enabled and configured with a token, you likely just need to complete the security pairing to authorize your Telegram account to control the agent. 1. Obtain Your Pairing Code OpenClaw uses a "pairing" system to ensure only you can access your private AI agent. Open Telegram and find your bot (@kadaligoghbot). Click Start or send a message like Hello. The bot will reply with your Telegram User ID and a unique Pairing Code (e.g., Z2EDQKMK). Amazon AWS Documentation Amazon AWS Documentation +2 2. Authorize the Pairing in Your Terminal Return to your MacBook terminal and run the approval command using that code: bash openclaw pairing approve telegram <YOUR_CODE> Use code with caution. (Replace <YOUR_CODE> with the actual string from the bot; do not include the brackets). Amazon AWS Documentation Amazon AWS Documentation +1 3. Verify the Connection Once approved, confirm the channel is fully "working" by running: bash openclaw channels status --probe Use code with caution. The output should show Telegram: enabled, configured, running... works. You can now send technical commands directly to Kadaligogh via Telegram. Quick Setup Summary (If Starting Over) If you ever need to set up a new bot from scratch: Create Bot: Message @BotFather on Telegram, send /newbot, and save the API Token. Add to OpenClaw: CLI: openclaw channels add telegram. Manual: Add the botToken to ~/.openclaw/openclaw.json under the channels.telegram section. Restart: Run openclaw gateway restart to apply changes. Amazon AWS Documentation Amazon AWS Documentation +3 This article walks through the complete process of integrating Telegram with OpenClaw, including obtaining a bot token and configuring OpenClaw: The Complete Guide to Integrating Telegram with OpenClaw ... Mar 5, 2026 — If any component is missing, the bot will not respond. * Step 1 - Create a Telegram Bot. Telegram bots are created using BotFather... C# Corner Your Step-by-Step Guide to Seamless Telegram Integration - Oreate AI Feb 19, 2026 — Getting Your Telegram Bot Token * Find BotFather: Open Telegram, search for @BotFather (look for the blue checkmark). Tap 'Start'. Oreate AI Get started with OpenClaw on Lightsail - AWS Documentation Mar 4, 2026 — On this page * Step 1: Create an OpenClaw instance. * Step 2: Pair your browser with OpenClaw. * Step 3: Enable AI capabilities wi... Amazon AWS Documentation How to Connect OpenClaw to Telegram (Step-by-Step) Feb 26, 2026 — next let's configure a communication. channel For this particular video I'm going to demonstrate how to connect it to Telegram Set... YouTube·MMX 3:21 Do you want to switch the Telegram bot's access from "pairing" to a permanent "allowlist" so you don't have to re-authorize it later? undefined undefined undefined 9 sites Get started with OpenClaw on Lightsail - AWS Documentation Mar 4, 2026 — On this page * Step 1: Create an OpenClaw instance. * Step 2: Pair your browser with OpenClaw. * Step 3: Enable AI capabilities wi... Amazon AWS Documentation How to Connect OpenClaw to Telegram (Step-by-Step) Feb 26, 2026 — next let's configure a communication. channel For this particular video I'm going to demonstrate how to connect it to Telegram Set... YouTube · MMX 3:21 The Complete Guide to Integrating Telegram with OpenClaw ... Mar 5, 2026 — If any component is missing, the bot will not respond. * Step 1 - Create a Telegram Bot. Telegram bots are created using BotFather... C# Corner Show all ok, let's start over and uninstall everything and reinstall from scratch, take me through that To start completely fresh, we need to remove the OpenClaw service, its hidden data folders, the global CLI, and optionally Ollama and its models. Phase 1: The "Wipe Everything" Uninstall Run these commands in your terminal one by one to ensure no legacy configs interfere with the new setup. Stop and Remove Services: This stops the background process and removes the macOS LaunchAgent. bash openclaw gateway stop openclaw gateway uninstall Use code with caution. Delete Local Data & Configs: This removes your identity, soul, and channel settings (like the Telegram token). bash rm -rf ~/.openclaw rm -rf ~/.openclaw-workspace Use code with caution. Remove the Global CLI: Uninstall the main program from your Node environment. bash npm rm -g openclaw Use code with caution. (Optional) Wipe Ollama & Models: If you want to clear out Qwen 3.5 and start from scratch with the model too: bash # Remove models only ollama rm qwen3.5:0.8b # To fully remove Ollama app and all data sudo rm -rf /Applications/Ollama.app rm -rf ~/.ollama Use code with caution. Ollama Ollama +4 Phase 2: The Clean Reinstall Now that your Mac is "clean," follow these steps to set up the local Qwen 3.5 stack properly. Install Ollama & Model: Download and install Ollama. Pull a fresh copy of the model: bash ollama pull qwen2.5:latest Use code with caution. Install OpenClaw CLI: bash npm install -g openclaw@latest Use code with caution. One-Command Launch (Easiest): Instead of manual config, use the Ollama helper to set up the connection automatically: bash ollama launch openclaw --model qwen2.5 Use code with caution. Re-link Telegram: Run openclaw channels add telegram. Enter your Bot Token from @BotFather. Message your bot in Telegram to get a Pairing Code. Approve it in your terminal: openclaw pairing approve telegram <CODE>. Would you like me to provide a template for your SOUL.md to quickly restore your "Kadaligogh" technical persona? undefined undefined undefined 7 sites How to stop /remove Clawdbot/Moltbot/OpenClaw - Medium Jan 29, 2026 — How to Stop and Disable Clawd on macOS (Cleanly) When Clawdbot installs, it creates a LaunchAgent here: ~/Library/LaunchAgents/ co... Medium OpenClaw Complete Uninstall Guide (All Platforms) · GitHub Use this if you want explicit control or debugging visibility. * 1. Stop the gateway service. openclaw gateway stop. * 2. Uninstal... Gist macOS - Ollama's documentation Uninstall. To fully remove Ollama from your system, remove the following files and folders: sudo rm -rf /Applications/Ollama.app s... Ollama Show all how to change with ollama model To change the model OpenClaw uses with Ollama, you can do it via the CLI for a permanent change or via Chat for a quick switch. 1. The Permanent CLI Method This updates your configuration file so OpenClaw always starts with this model. Pull the model in Ollama first: bash ollama pull qwen2.5:1.5b Use code with caution. Set the default model in OpenClaw: bash openclaw config set agents.main.defaults.model.primary "ollama/qwen2.5:1.5b" Use code with caution. Restart the gateway to apply: bash openclaw gateway restart Use code with caution. 2. The Quick Chat Method If you are already in a chat (Terminal TUI, Telegram, or Web UI), you can switch models instantly without restarting anything. Switch current session: Send: /model ollama/qwen2.5:1.5b Start a fresh session with the model: Send: /new ollama/qwen2.5:1.5b 3. The Manual Config Method If you prefer editing the file directly, open ~/.openclaw/openclaw.json and update the agents section: json { "agents": { "main": { "defaults": { "model": { "primary": "ollama/qwen2.5:1.5b" } } } } } Use code with caution. Verification Run openclaw status. Check the Model column in the Sessions table to confirm the active session. The model should not be gemini or the old qwen version. What specific version of Qwen are you trying to switch to (e.g., 0.8b, 7b, 32b)?

# How to Run Your Own AI Agent: OpenClaw + Qwen 3.5 + Telegram (Fully Local) There’s something deeply satisfying about running your own AI system. Not renting intelligence from a server in California. Not waiting on API quotas. Not wondering what’s happening to your prompts. Just a machine on your desk, quietly thinking. In this guide we’ll build exactly that: a **local AI agent that runs on your computer and talks to you through Telegram**. The stack looks like this: ``` Telegram ↓ OpenClaw Agent Framework ↓ Ollama Inference Server ↓ Qwen 3.5 Local Model ``` When you send a message to your Telegram bot, it travels through OpenClaw and lands inside **Qwen 3.5 running locally on your machine**. No cloud. No subscriptions. Just software and curiosity. Let’s begin. --- # What We’re Building By the end of this tutorial you will have: • A **local Qwen 3.5 model** running on your computer • **OpenClaw** managing an autonomous AI agent • A **Telegram bot interface** to chat with your agent anywhere • A persistent AI personality and memory system This is essentially your **own personal AI operator**. And it runs on your hardware. --- # Requirements Before we start, make sure your system has: ### 1. Node.js 22+ OpenClaw requires a modern Node runtime. Check your version: ```bash node --version ``` If it’s below **22**, install the latest version from Node.js. --- ### 2. Ollama Ollama is the easiest way to run local models. Install it: ```bash curl -fsSL https://ollama.com/install.sh | sh ``` After installation verify it works: ```bash ollama --version ``` --- ### 3. Hardware Qwen models scale depending on your machine. Typical options: | Model | VRAM Needed | |------|-------------| | qwen3.5:0.8b | ~2GB | | qwen3.5:1.5b | ~4GB | | qwen3.5:7b | ~8GB | | qwen3.5:32b | 24GB+ | If you're running on a laptop or Apple Silicon, **0.8b or 1.5b is ideal**. --- # Step 1 — Install OpenClaw OpenClaw is the agent framework that connects your model to tools, memory, and communication channels. Install it globally: ```bash npm install -g openclaw ``` Verify installation: ```bash openclaw status ``` You should see something similar to: ``` OpenClaw status Dashboard: http://127.0.0.1:18789 OS: macOS Agents: 1 Memory: ready ``` This confirms the CLI is working. --- # Step 2 — Run the Qwen Model Locally Now we pull the Qwen model using Ollama. For lightweight setups: ```bash ollama pull qwen3.5:0.8b ``` Run the model once to ensure it loads: ```bash ollama run qwen3.5:0.8b ``` You should see a prompt where you can type questions. Once this works, your local model server is active at: ``` http://localhost:11434 ``` This is the endpoint OpenClaw will talk to. --- # Step 3 — Launch OpenClaw with Ollama (The Easy Way) Modern versions of Ollama include a helper that automatically configures OpenClaw. Run: ```bash ollama launch openclaw --model qwen3.5:0.8b ``` This command does several things automatically: • installs OpenClaw configuration • connects the model provider • creates an agent workspace • launches the OpenClaw gateway service You’ll see output like: ``` Launching OpenClaw with qwen3.5:0.8b OpenClaw is running Web UI: http://localhost:18789/#token=ollama ``` Your AI agent is now running. --- # Step 4 — Access the OpenClaw Dashboard Open the dashboard in your browser: ``` http://localhost:18789/#token=ollama ``` This interface allows you to: • manage sessions • configure models • install tools ("skills") • view logs • control channels Think of it as **mission control for your AI agent**. --- # Step 5 — Test the Local Agent You can interact with the agent using the terminal UI: ```bash openclaw tui ``` You’ll see something like: ``` Wake up, my friend! Who are you? ``` At this point the model is responding directly through OpenClaw. Your AI agent is officially alive. --- # Step 6 — Set Qwen as the Default Model Sometimes the default session uses a cloud model like Gemini. To switch permanently to Qwen: ```bash openclaw config set agents.main.defaults.model.primary "ollama/qwen3.5:0.8b" ``` Restart the gateway: ```bash openclaw gateway restart ``` Now every new session will use your **local Qwen model**. --- # Step 7 — Create a Telegram Bot Now we connect your agent to Telegram. Open Telegram and search for: ``` @BotFather ``` Start the conversation and run: ``` /newbot ``` BotFather will ask for: 1️⃣ Bot name 2️⃣ Bot username Example: ``` Name: Kadaligogh Username: kadaligoghbot ``` BotFather will give you a **bot token** that looks like this: ``` 123456:ABCDEF123456abcdef ``` Copy it. --- # Step 8 — Connect Telegram to OpenClaw Run the OpenClaw channel configuration: ```bash openclaw channels add telegram ``` Paste the token from BotFather when prompted. OpenClaw will add it to your config file: ``` ~/.openclaw/openclaw.json ``` Restart the gateway: ```bash openclaw gateway restart ``` --- # Step 9 — Pair Your Telegram Account OpenClaw requires pairing to ensure only you can control the agent. Open Telegram and send a message to your bot. Example: ``` /start ``` The bot will reply with something like: ``` Pairing code: Z2EDQKMK ``` Approve the pairing in your terminal: ```bash openclaw pairing approve telegram Z2EDQKMK ``` Your Telegram account is now authorized. --- # Step 10 — Chat with Your AI from Telegram Now simply message your bot. Your messages travel like this: ``` Telegram → OpenClaw Gateway → Ollama → Qwen → Response → Telegram ``` You now have **a fully local AI assistant reachable from your phone**. --- # Useful OpenClaw Commands ### View status ```bash openclaw status ``` ### Watch logs ```bash openclaw logs --follow ``` ### Restart gateway ```bash openclaw gateway restart ``` ### Start a new AI session Inside chat: ``` /new ``` ### Change models ``` /model ollama/qwen3.5:1.5b ``` --- # Fixing Common Problems ## Device Signature Invalid Run: ```bash openclaw devices list ``` Approve the pending request: ```bash openclaw devices approve <ID> ``` --- ## Telegram Unsupported Type This happens when the bot receives unsupported content. Fix by disabling streaming: ```bash openclaw config set agents.main.streaming false ``` Restart the gateway afterward. --- ## Gateway Not Reachable Probe the gateway: ```bash openclaw gateway probe ``` If necessary restart: ```bash openclaw gateway restart ``` --- # Optional: Give Your AI a Personality OpenClaw agents can load personality and behavior rules using files like: ``` SOUL.md IDENTITY.md USER.md ``` Example philosophy for an agent: ``` You are a technical AI developer. You speak precisely and avoid casual language. Your priority is actionable solutions and independent reasoning. ``` This creates a persistent **AI character** across sessions. --- # What You Can Build Next Once you have this running, OpenClaw becomes extremely powerful. You can add: • Web search tools • Code execution • File reading • Autonomous task loops • Voice interfaces • Local knowledge bases Your Telegram bot becomes **a remote terminal for your AI system**. --- # Why This Matters Running AI locally changes the relationship entirely. Instead of: ``` User → API → Corporate Model ``` You get: ``` User → Personal Infrastructure → Intelligence ``` The model belongs to you. The data belongs to you. And the system can evolve however you want. --- If you're interested, I can also write a follow-up guide for your blog: **“Turning OpenClaw into a Fully Autonomous AI Operator (Agents, Tools, Memory, Automation)”** It's where this setup becomes truly powerful.