Skip to content

[Bug]: The local ollama llm was invalid after 4.26 version #76117

@MarsCNS

Description

@MarsCNS

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

When I updated the vsesion to v2026.4.29, the local ollama LLM was ruined. Before v2026.4.26, the local LLM thinks slowly, but it is precious call the tools, knowing your purpose. And now, the local LLM don't know my commands, but the expansive remote LLM can. I asked these LLM, they said the codes, like selection-CwAy0mf2.js, limits the timeout seconds to 120s.
for example: const DEFAULT_LLM_IDLE_TIMEOUT_MS = 120 * 1e3;(line 5977)
When local LLM is thinking, they cut off it and let the llm says nonsense.

So I fixed it, and the local LLM still act like a moron. Too many codes to review.

I wish you can fix the issues because I don't care about the latency. I can wait the more precise answer. Thank you.

Steps to reproduce

  1. Updated the openclaw to v2026.4.29 without the openclaw.json changing.
  2. Ask the ollama/qwen3.6:latest.

Expected behavior

Using correct tools and good answer.

Actual behavior

says nonsense.

OpenClaw version

2026.4.29

Operating system

Ubuntu 24

Install method

No response

Model

ollama/qwen3.6:35B:a3b

Provider / routing chain

gateway webchat

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Impact and severity

No response

Additional information

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingregressionBehavior that previously worked and now fails

    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