issue: OWUI cannot parse tool call parameters when LLM makes multiple tool calls at once #5709

New Issue

2025-11-11T16:30:51-06:00

GiteaMirror commented

2025-11-11 16:30:51 -06:00

Originally created by @mlaihk on GitHub (Jul 6, 2025).

Check Existing Issues

I have searched the existing issues and discussions.
I am using the latest version of Open WebUI.

Installation Method

Docker

Open WebUI Version

0.6.15

Ollama Version (if applicable)

0.95

Operating System

Windows 11 24H2

Browser (if applicable)

MSEdge

Confirmation

I have read and followed all instructions in README.md.
I am using the latest version of both Open WebUI and Ollama.
I have included the browser console logs.
I have included the Docker container logs.
I have provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.

Expected Behavior

When LLM makes multiple tool calls at a iteration, Open WebUI should correctly parse the calls into multiple tool calls and execute.

Actual Behavior

Open WebUI fails to parse the multiple tool calls by the LLM and throws an error.

The docker container contains this error directly related to the multiple tool call.
2025-07-06 11:18:31.178 | ERROR | open_webui.utils.middleware:post_response_handler:2080 - Error parsing tool call arguments: {"include_urls": true, "url": "https://rog.asus.com/hk-en/laptops/rog-zephyrus/rog-zephyrus-g14-2025-ga403um-qs013w/"}{"include_urls": true, "url": "https://rog.asus.com/hk-en/laptops/rog-zephyrus/rog-zephyrus-g16-2025-gu605/"} - {}

Steps to Reproduce

Use Ollama hosted Qwen3:14b or Gemma3:12b. As LLM a question that will result in multiple tool calls:

Example:
use playwright and access links from the search and assemble complete specifications on all the 2025 Zephyrus G14 and G16 available. Make sure the models are 2025 models, with the correct CPUs and dGPUs.

Logs & Screenshots

As above

Additional Information

No response

Originally created by @mlaihk on GitHub (Jul 6, 2025). ### Check Existing Issues - [x] I have searched the existing issues and discussions. - [x] I am using the latest version of Open WebUI. ### Installation Method Docker ### Open WebUI Version 0.6.15 ### Ollama Version (if applicable) 0.95 ### Operating System Windows 11 24H2 ### Browser (if applicable) MSEdge ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have included the browser console logs. - [x] I have included the Docker container logs. - [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.** - [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc). - [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps: - Start with the initial platform/version/OS and dependencies used, - Specify exact install/launch/configure commands, - List URLs visited, user input (incl. example values/emails/passwords if needed), - Describe all options and toggles enabled or changed, - Include any files or environmental changes, - Identify the expected and actual result at each stage, - Ensure any reasonably skilled user can follow and hit the same issue. ### Expected Behavior When LLM makes multiple tool calls at a iteration, Open WebUI should correctly parse the calls into multiple tool calls and execute. ### Actual Behavior Open WebUI fails to parse the multiple tool calls by the LLM and throws an error. <img width="1065" height="292" alt="Image" src="https://github.com/user-attachments/assets/fce544b9-8073-43a6-8e7d-1dfd7a3186b6" /> The docker container contains this error directly related to the multiple tool call. `2025-07-06 11:18:31.178 | ERROR | open_webui.utils.middleware:post_response_handler:2080 - Error parsing tool call arguments: {"include_urls": true, "url": "https://rog.asus.com/hk-en/laptops/rog-zephyrus/rog-zephyrus-g14-2025-ga403um-qs013w/"}{"include_urls": true, "url": "https://rog.asus.com/hk-en/laptops/rog-zephyrus/rog-zephyrus-g16-2025-gu605/"} - {}` ### Steps to Reproduce Use Ollama hosted Qwen3:14b or Gemma3:12b. As LLM a question that will result in multiple tool calls: Example: use playwright and access links from the search and assemble complete specifications on all the 2025 Zephyrus G14 and G16 available. Make sure the models are 2025 models, with the correct CPUs and dGPUs. ### Logs & Screenshots As above ### Additional Information _No response_

GiteaMirror added the bug label 2025-11-11 16:30:51 -06:00

GiteaMirror closed this issue

2025-11-11 16:30:52 -06:00

GiteaMirror commented

2025-11-11 16:30:54 -06:00

@tjbck commented on GitHub (Jul 6, 2025):

Local models are not reliable for tool calling especially at <14b size. I'd encourage you to use higher-end models like gpt-4.1 for such tasks.

@tjbck commented on GitHub (Jul 6, 2025): Local models are not reliable for tool calling especially at <14b size. I'd encourage you to use higher-end models like gpt-4.1 for such tasks.

GiteaMirror referenced this issue

2026-04-19 20:34:25 -05:00

[GH-ISSUE #5709] Upgrade using pip install caused settings to default back to normal after server restart #14095

GiteaMirror referenced this issue

2026-04-25 03:59:11 -05:00

[GH-ISSUE #5709] Upgrade using pip install caused settings to default back to normal after server restart #29623

GiteaMirror referenced this issue

2026-05-05 13:52:31 -05:00

[GH-ISSUE #5709] Upgrade using pip install caused settings to default back to normal after server restart #52761

GiteaMirror referenced this issue

2026-05-13 01:03:16 -05:00

[GH-ISSUE #5709] Upgrade using pip install caused settings to default back to normal after server restart #68699