Issue Station
Sign In
Tools hereToolbox

openclaw - 💡(How to fix) Fix [Bug:] Dashboard and CLI fail to render responses with Ollama integration (Qwen2.5) [1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#57274•Fetched 2026-04-08 01:51:44
View on GitHub
Comments
0
Participants
1
Timeline
2
Reactions
0
Author
ClovisChProgrammerClovisChProgrammer
Participants
ClovisChProgrammerClovisChProgrammer
Timeline (top)
labeled ×2

Troubleshooting Report: OpenClaw + Ollama (Qwen2.5) Context Environment: Windows 10/11, PowerShell

Gateway Command: openclaw gateway --port 18789

Target Model: ollama/qwen2.5:latest

Goal: Use OpenClaw’s dashboard to interact with Qwen2.5 running locally via Ollama

Problems and Solutions Problem 1: Incorrect Fallback to Anthropic Symptom: Gateway log shows fallback to anthropic/qwen2.5:latest.

Error: Unknown model: anthropic/qwen2.5:latest.

Impact: Model not found, agent fails before reply.

Solution Attempted: Explicitly set "primary": "ollama/qwen2.5:latest".

Result: Gateway correctly recognized Ollama model, but dashboard still failed to display responses.

Problem 2: Invalid Key in Configuration Symptom: Config error: agents: Unrecognized key: "entries".

Impact: Gateway refused to start with invalid config.

Solution Attempted: Removed "entries" key from agents.

Result: Configuration accepted, gateway initialized.

Problem 3: Invalid API Field Symptom: Config error: models.providers.ollama.api: Invalid input.

Impact: Gateway refused to run when "api": "openai-chat" was used.

Solution Attempted: Changed "api" to "openai-completions".

Result: Configuration valid, but dashboard did not render responses.

Problem 4: Dashboard Does Not Display Responses Symptom: Chat input blinks, no text shown.

Log: embedded run done ... aborted=false.

Impact: Ollama responded, but output not rendered in dashboard.

Solution Attempted: Multiple config adjustments, including API field changes.

Result: No improvement; issue appears to be a UI rendering bug.

Problem 5: CLI Agent Command Errors Symptom:

Error: Pass --to <E.164>, --session-id, or --agent to choose a session

HTTP 404: 404 page not found

Impact: CLI unable to send messages without valid session/channel.

Solution Attempted: Used --agent main and --session-id test.

Result: Still returned 404; CLI requires external channel configuration.

Conclusions Ollama is fully functional and responds correctly when tested directly (ollama run qwen2.5:latest).

OpenClaw recognizes Ollama but fails to display responses in both dashboard and CLI.

Issues are caused by integration layer bugs: fallback misrouting, invalid API field handling, and missing response rendering.

The problem is not with Ollama, but with OpenClaw’s handling of local model responses.

Recommendations for Developers Fix Fallback Behavior

Prevent fallback to anthropic/... when ollama/... is specified.

Ensure provider prefix is respected.

Support Ollama Response Format

Adjust parser to correctly display responses in dashboard when using "openai-completions".

Consider adding official support for "openai-chat".

Improve CLI Agent Command

Allow direct messaging to default agent without requiring --session-id or external channels.

Avoid 404 errors when no channel is configured.

Enhance Documentation

Clarify accepted values for "api".

Provide working examples for Ollama integration.

Final Summary: Ollama works perfectly, but OpenClaw fails to display responses in the dashboard and CLI due to integration issues. The report highlights fallback errors, invalid config keys, and rendering problems, with actionable recommendations for developers.

Error Message

Logs, Screenshots, and Evidence

Gateway log when provider prefix is missing: Model "qwen2.5:latest" specified without provider. Falling back to "anthropic/qwen2.5:latest". Error: Unknown model: anthropic/qwen2.5:latest

Configuration error with invalid key: Invalid config at openclaw.json: agents: Unrecognized key: "entries"

Configuration error with API field: models.providers.ollama.api: Invalid input

Dashboard symptom: Chat input blinks, no response text displayed. Log: embedded run done ... aborted=false

CLI errors: Error: Pass --to <E.164>, --session-id, or --agent to choose a session HTTP 404: 404 page not found

Direct Ollama test: Running ollama run qwen2.5:latest returns correct responses, confirming the issue is not with Ollama itself.

Root Cause

Issues are caused by integration layer bugs: fallback misrouting, invalid API field handling, and missing response rendering.

Code Example

Logs, Screenshots, and Evidence

Gateway log when provider prefix is missing:
Model "qwen2.5:latest" specified without provider. Falling back to "anthropic/qwen2.5:latest".
Error: Unknown model: anthropic/qwen2.5:latest

Configuration error with invalid key:
Invalid config at openclaw.json: agents: Unrecognized key: "entries"

Configuration error with API field:
models.providers.ollama.api: Invalid input

Dashboard symptom:
Chat input blinks, no response text displayed.
Log: embedded run done ... aborted=false

CLI errors:
Error: Pass --to <E.164>, --session-id, or --agent to choose a session
HTTP 404: 404 page not found

Direct Ollama test:
Running ollama run qwen2.5:latest returns correct responses, confirming the issue is not with Ollama itself.
RAW_BUFFERClick to expand / collapse

Bug type

Crash (process/app exits or hangs)

Beta release blocker

Yes

Summary

Troubleshooting Report: OpenClaw + Ollama (Qwen2.5) Context Environment: Windows 10/11, PowerShell

Gateway Command: openclaw gateway --port 18789

Target Model: ollama/qwen2.5:latest

Goal: Use OpenClaw’s dashboard to interact with Qwen2.5 running locally via Ollama

Problems and Solutions Problem 1: Incorrect Fallback to Anthropic Symptom: Gateway log shows fallback to anthropic/qwen2.5:latest.

Error: Unknown model: anthropic/qwen2.5:latest.

Impact: Model not found, agent fails before reply.

Solution Attempted: Explicitly set "primary": "ollama/qwen2.5:latest".

Result: Gateway correctly recognized Ollama model, but dashboard still failed to display responses.

Problem 2: Invalid Key in Configuration Symptom: Config error: agents: Unrecognized key: "entries".

Impact: Gateway refused to start with invalid config.

Solution Attempted: Removed "entries" key from agents.

Result: Configuration accepted, gateway initialized.

Problem 3: Invalid API Field Symptom: Config error: models.providers.ollama.api: Invalid input.

Impact: Gateway refused to run when "api": "openai-chat" was used.

Solution Attempted: Changed "api" to "openai-completions".

Result: Configuration valid, but dashboard did not render responses.

Problem 4: Dashboard Does Not Display Responses Symptom: Chat input blinks, no text shown.

Log: embedded run done ... aborted=false.

Impact: Ollama responded, but output not rendered in dashboard.

Solution Attempted: Multiple config adjustments, including API field changes.

Result: No improvement; issue appears to be a UI rendering bug.

Problem 5: CLI Agent Command Errors Symptom:

Error: Pass --to <E.164>, --session-id, or --agent to choose a session

HTTP 404: 404 page not found

Impact: CLI unable to send messages without valid session/channel.

Solution Attempted: Used --agent main and --session-id test.

Result: Still returned 404; CLI requires external channel configuration.

Conclusions Ollama is fully functional and responds correctly when tested directly (ollama run qwen2.5:latest).

OpenClaw recognizes Ollama but fails to display responses in both dashboard and CLI.

Issues are caused by integration layer bugs: fallback misrouting, invalid API field handling, and missing response rendering.

The problem is not with Ollama, but with OpenClaw’s handling of local model responses.

Recommendations for Developers Fix Fallback Behavior

Prevent fallback to anthropic/... when ollama/... is specified.

Ensure provider prefix is respected.

Support Ollama Response Format

Adjust parser to correctly display responses in dashboard when using "openai-completions".

Consider adding official support for "openai-chat".

Improve CLI Agent Command

Allow direct messaging to default agent without requiring --session-id or external channels.

Avoid 404 errors when no channel is configured.

Enhance Documentation

Clarify accepted values for "api".

Provide working examples for Ollama integration.

Final Summary: Ollama works perfectly, but OpenClaw fails to display responses in the dashboard and CLI due to integration issues. The report highlights fallback errors, invalid config keys, and rendering problems, with actionable recommendations for developers.

Steps to reproduce

Steps to Reproduce

Install OpenClaw on Windows (tested on Windows 10/11).

Configure the file openclaw.json with Ollama provider and model qwen2.5:latest.

Start the gateway using the command: openclaw gateway --port 18789.

Open the dashboard and attempt to send a message to the agent.

Observe that the chat input blinks but no response text is displayed.

Run the CLI command: openclaw agent --local --agent main --message "Hello".

Observe that the CLI returns HTTP 404 or requires a session/channel, preventing any response from being shown.

Run Ollama directly with the command ollama run qwen2.5:latest and confirm that the model responds correctly outside of OpenClaw.

Expected behavior

Expected Behavior

When sending a message through the dashboard, the response from the Ollama model (qwen2.5:latest) should be displayed clearly in the chat window.

The CLI command openclaw agent --local --agent main --message "Hello" should return the model’s response directly in the terminal without requiring extra session or channel configuration.

OpenClaw should correctly parse and render the output from Ollama using the configured API field, without falling back to unsupported providers or returning HTTP 404 errors.

Actual behavior

Actual Behavior

When sending a message through the dashboard, the input field blinks but no response text is displayed.

Gateway logs show that Ollama responds, but the output is not rendered in the dashboard.

When running the CLI command openclaw agent --local --agent main --message "Hello", the system returns HTTP 404 or requires a session/channel, preventing any response from being shown.

Attempts to use --session-id still result in 404 errors.

Configuration changes such as setting "api": "openai-chat" are rejected as invalid input.

Running Ollama directly (ollama run qwen2.5:latest) works correctly, confirming that the issue is not with Ollama but with OpenClaw’s integration and rendering.

OpenClaw version

OpenClaw Version OpenClaw 2026.1.29 (commit a5b4d22) Running on Windows 10/11 via PowerShell

Operating system

Windows 10

Install method

Normal.

Model

Qwen2.5:Latest

Provider / routing chain

Provider / Routing Chain Provider: Ollama Model: qwen2.5:latest Routing chain: OpenClaw → Ollama (local) → Qwen2.5 response Note: When provider prefix is missing, OpenClaw incorrectly falls back to anthropic/qwen2.5:latest, which does not exist.

Additional provider/model setup details

Provider / Routing Chain

Provider: Ollama

Model: qwen2.5:latest

Routing chain: OpenClaw → Ollama (local) → Qwen2.5 response

Note: When provider prefix is missing, OpenClaw incorrectly falls back to anthropic/qwen2.5:latest, which does not exist.

Logs, screenshots, and evidence

Logs, Screenshots, and Evidence

Gateway log when provider prefix is missing:
Model "qwen2.5:latest" specified without provider. Falling back to "anthropic/qwen2.5:latest".
Error: Unknown model: anthropic/qwen2.5:latest

Configuration error with invalid key:
Invalid config at openclaw.json: agents: Unrecognized key: "entries"

Configuration error with API field:
models.providers.ollama.api: Invalid input

Dashboard symptom:
Chat input blinks, no response text displayed.
Log: embedded run done ... aborted=false

CLI errors:
Error: Pass --to <E.164>, --session-id, or --agent to choose a session
HTTP 404: 404 page not found

Direct Ollama test:
Running ollama run qwen2.5:latest returns correct responses, confirming the issue is not with Ollama itself.

Impact and severity

No response

Additional information

No response

extent analysis

Fix Plan

To resolve the issues with OpenClaw's integration with Ollama, follow these steps:

  1. Fix Fallback Behavior:
    • Ensure the provider prefix is respected by modifying the openclaw.json configuration file to include the provider name.
    • Example: 
      {
  "primary": "ollama/qwen2.5:latest"
}
  2. Support Ollama Response Format:
    • Adjust the parser to correctly display responses in the dashboard when using "openai-completions".
    • Modify the API field in openclaw.json to use "openai-completions".
    • Example: 
      {
  "models": {
    "providers": {
      "ollama": {
        "api": "openai-completions"
      }
    }
  }
}
  3. Improve CLI Agent Command:
    • Update the CLI command to allow direct messaging to the default agent without requiring --session-id or external channels.
    • Modify the openclaw agent command to use the --local and --agent flags.
    • Example: 
      openclaw agent --local --agent main --message "Hello"
  4. Enhance Documentation:
    • Clarify accepted values for the "api" field in the documentation.
    • Provide working examples for Ollama integration in the documentation.

Code Changes

To implement the fixes, update the OpenClaw codebase with the following changes:

  • In openclaw/gateway.py, add a check to respect the provider prefix: 
    if "provider" in config:
    model_name = f"{config['provider']}/{config['model']}"
  • In openclaw/parser.py, update the parser to support the Ollama response format: 
    if config["api"] == "openai-completions":
    # Parse response using openai-completions format
    pass
  • In openclaw/cli.py, update the CLI command to allow direct messaging to the default agent: 
    if args.local and args.agent:
    # Send message to default agent
    pass

Verification

To verify the fixes, follow these steps:

  1. Restart the OpenClaw gateway with the updated configuration.
  2. Send a message through the dashboard and verify that the response is displayed correctly.
  3. Run the CLI command with the --local and --agent flags and verify that the response is returned correctly.

Extra Tips

  • Ensure that the Ollama model is correctly configured and responding as expected.
  • Test the fixes with different provider prefixes and API fields to ensure compatibility.
  • Update the documentation to reflect the changes and provide working examples

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

FAQ

Expected behavior

Expected Behavior

When sending a message through the dashboard, the response from the Ollama model (qwen2.5:latest) should be displayed clearly in the chat window.

The CLI command openclaw agent --local --agent main --message "Hello" should return the model’s response directly in the terminal without requiring extra session or channel configuration.

OpenClaw should correctly parse and render the output from Ollama using the configured API field, without falling back to unsupported providers or returning HTTP 404 errors.

#api#dependency error#configuration error#environment variable#integration issue

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix [Bug:] Dashboard and CLI fail to render responses with Ollama integration (Qwen2.5) [1 participants]