Skip to content

Claw is ~120s slower than direct Ollama/Qwen2.5:7b calls on a local setup #3252

@requiem-2249

Description

@requiem-2249

Claw is configured for a local Ollama-backed qwen2.5:7b setup on Linux Mint. It loads both ~/.claw/settings.json and /home/prakhar/claw-code/rust/.claw.json, resolves the model as qwen2.5:7b, and uses workspace-write permissions. The shell environment contains:

OPENAI_BASE_URL=http://127.0.0.1:11434/v1

Ollama itself appears healthy and local:

  • curl -v http://127.0.0.1:11434/api/tags returns HTTP 200 and shows qwen2.5:7b.
  • ss -tlnp confirms Ollama is listening on 127.0.0.1:11434.
  • A direct OpenAI-compatible request to http://127.0.0.1:11434/v1/chat/completions completes in approximately 7.4s for a trivial prompt.

However, Claw interactive prompts such as:

  • Hi
  • How are you?

take approximately 2 minutes to return a response.

Additional observations:

  • The issue reproduces even when Claw is started from an unrelated directory outside the repository.
  • strace shows Claw connecting directly to 127.0.0.1:11434.
  • No connections to external providers were observed during testing.
  • Direct requests to the same local Ollama endpoint complete in approximately 7 seconds, while equivalent prompts through Claw take approximately 120 seconds.

Attached files:

  1. claw.txt - strace output captured during a slow interaction.
  2. Report.txt - additional diagnostics and command output.

Please let me know if there are additional diagnostics I should collect.

Report.txt

claw.txt

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions