🦞OpenClaw Guide
Models

Model Not Responding — No Output or Empty Replies

Troubleshooting guide for when OpenClaw shows "(no output)" or produces empty responses. Covers authentication issues, gateway problems, model configuration, and platform-specific fixes for VPS, WSL2, and Docker.

⚠️ The Problem

OpenClaw appears to be running but produces no response when you send messages. Common symptoms include:

  • TUI displays (no output) after sending a message
  • Status shows "connected | idle" but nothing happens
  • Token counter stays at 0/200k (0%) indefinitely
  • Gateway logs show embedded run agent start and embedded run agent end at the same moment with no errors
  • Model responds via dashboard but not TUI
  • Works with one provider (e.g., ChatGPT) but not another (e.g., Claude)

🔍 Why This Happens

The "(no output)" message appears when both finalText and streamedText are empty — meaning the model didn't produce any response. This typically happens due to:

  1. Authentication failure: Invalid API key, expired OAuth token, or missing credentials. Error: HTTP 401 authentication_error: Invalid bearer token (request_id: req_...)

  2. Gateway not running: The agent can't communicate without an active gateway service.

  3. Model not configured: No default model set, or the configured model isn't accessible.

  4. API key not loading: Environment variables not inherited by the daemon, especially common in WSL2 and VPS setups.

  5. OAuth token host-specific: Claude CLI OAuth tokens are tied to the machine where they were generated — they don't work on VPS/remote servers.

  6. Version mismatch: Running an outdated version with known bugs.

  7. Allowlist blocking: Sender not on the allowlist, so messages are ignored.

✅ The Fix

Quick Triage (First 60 Seconds)

Run these commands in order to identify the issue:

bash
# Check overall statusopenclaw status# Check model configuration and authopenclaw models status# Watch live logs for errorsopenclaw gateway logs --follow

Fix 1: Gateway Not Running

If openclaw status shows the gateway is stopped:

bash
# Check gateway statusopenclaw gateway status# Start or restart the gatewayopenclaw gateway restart# For VPS: ensure daemon is installedopenclaw gateway install --install-daemon

Fix 2: Authentication Issues (HTTP 401)

If you see HTTP 401 authentication_error: Invalid bearer token:

bash
# Check which auth profile is activeopenclaw models status# Re-sync Claude CLI token (if using OAuth)claude setup-token# OR via OpenClaw:openclaw models auth setup-token --provider anthropic

For VPS/remote servers, use API key instead of OAuth (more reliable):

bash
# Add to ~/.openclaw/.envecho 'ANTHROPIC_API_KEY=sk-ant-api03-...' >> ~/.openclaw/.env# Restart to applyopenclaw gateway restart

Important: OAuth tokens are host-specific. If you set up on your laptop, those tokens won't work on your VPS. Always use API keys for remote servers.

Fix 3: Model Not Configured

If openclaw models status shows no credentials or no default model:

bash
# Run the configuration wizardopenclaw onboard# Or manually check your configopenclaw models list

Make sure your config has a valid default model:

json
{  "agents": {    "defaults": {      "model": {        "primary": "anthropic/claude-sonnet-4-5"      }    }  }}

Fix 4: API Key Not Loading (WSL2/VPS)

Environment variables from your shell aren't inherited by the daemon. Add them to the .env file:

bash
# Create or edit the .env filenano ~/.openclaw/.env# Add your API keysANTHROPIC_API_KEY=sk-ant-api03-...OPENAI_API_KEY=sk-...# Restart gateway to load new envopenclaw gateway restart

Fix 5: Version Mismatch

Version mismatches are the #1 cause of weird issues. Always check your version:

bash
# Check current versionopenclaw --version# Update to latestpnpm update -g openclaw# ORbrew update && brew upgrade openclaw# ORscoop update openclaw# Restart after updateopenclaw gateway restart

Fix 6: Run the Doctor

The built-in doctor command can auto-fix many issues:

bash
openclaw doctor --fix

This checks and repairs:

  • Gateway connectivity
  • Model configuration
  • Auth credentials
  • Channel connections

Fix 7: Check Allowlist Settings

If messages are being ignored, check the allowlist:

bash
# Check allowlist statusopenclaw status | grep -i allow# View full configopenclaw config get

Fix 8: Enable Delivery

In the TUI, make sure message delivery is enabled:

bash
# In the TUI, run:/status/deliver on

Fix 9: Check Channel Health

Verify your messaging platform is properly connected:

bash
# Check all channel statusesopenclaw channels status

Debugging Workflow

If the above fixes don't work, follow this debugging workflow:

bash
# 1. Get detailed statusopenclaw status --deep# 2. Check logs for specific errorsopenclaw gateway logs --follow# 3. Test model directlyopenclaw test# 4. Verify channel connectionopenclaw channels status# 5. Full diagnostic with auto-repairopenclaw doctor --fix

Platform-Specific Notes

VPS/Remote Servers

  • Always use API keys, not OAuth tokens
  • Ensure port 18789 is accessible
  • Run openclaw gateway install --install-daemon for persistent service

WSL2 (Windows)

  • Put API keys in ~/.openclaw/.env, not shell exports
  • Try both localhost and 127.0.0.1 for Ollama connections
  • Restart WSL2 if networking issues persist: wsl --shutdown

Docker

  • Use host.docker.internal instead of 127.0.0.1 for host services
  • Set env vars in docker-compose.yml, not shell exports

Still Stuck?

If none of these fixes work, share the following in the #help channel:

  1. Output of openclaw --version
  2. Output of openclaw models status
  3. Relevant lines from openclaw gateway logs
  4. Your platform (macOS/Linux/Windows WSL2/Docker)

🚀 Skip the setup. OpenClaw Cloud is ready in 60 seconds.

No server. No SSH. No config files. Just connect your channel and go.

Deploy Now — 50% Off First Month →

🔥 Your AI should run your business, not just answer questions.

We'll show you how.Free to join.

Join Vibe Combinator →

📋 Quick Commands

CommandDescription
openclaw statusCheck overall system status including gateway and connections
openclaw models statusVerify model configuration and authentication status
openclaw gateway logs --followStream live gateway logs to see errors
openclaw gateway restartRestart the gateway service
openclaw doctor --fixRun diagnostics and auto-repair common issues
openclaw onboardRe-run the configuration wizard
openclaw --versionCheck current installed version
openclaw testSend a test message to verify model responds
openclaw channels statusCheck messaging platform connections
openclaw status --deepDetailed status including model probe results

Related Issues

📚 You Might Also Like

Guide

How to Change AI Model in OpenClaw (Claude, GPT-4, Gemini)

Switch AI models in OpenClaw with /model or edit openclaw.json. Supports Claude Opus, Sonnet, GPT-4, Gemini, MiniMax, and 20+ models. Change per-session or set a default.

Blog

OpenClaw Group Chat Security: Why Your Bot Should Never Be in a Group

Here's the threat model almost nobody thinks about when setting up OpenClaw: if your bot is in a group chat, everyone in that group can give it commands.

Comparison

OpenClaw vs Notion AI

Your assistant, everywhere - not just in one app

Use Case

AI Assistant for Students

Study smarter, not harder

🐙 Your AI should run your business.

Weekly live builds + template vault. We'll show you how to make AI actually work.Free to join.

Join Vibe Combinator →