github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12690] Web search does not work properly with non gpt-oss models #70477

New Issue
Open
opened 2026-05-04 21:40:51 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-05-04 21:40:51 -05:00
Owner
Copy Link

Originally created by @ghost on GitHub (Oct 18, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12690

What is the issue?

Version 0.12.6 adds web search in the GUI to tool-calling models like Qwen3, but then it returns 102,207 tokens for a simple query, the same query using gpt-oss:20b-cloud web search returns 768 tokens.

My test prompt was "Who is the current US president search" and it seems like for Qwen3 it just returns 5 webpages and the entire contents of each, so one of them was the Wikipedia page for US president, which is about 100,000 tokens.

For gpt-oss:20b-cloud the formatting of web search results is completely different however, just returning a small 768 token snippet. Is this because qwen3:4b sucks at calling ollama's web search properly?

In Ollama's blog post (https://ollama.com/blog/web-search), they do say it can return thousands of tokens and to set context length to at least 32k, but still, 102,207 tokens is a bit much. Any reason for this?

Relevant log output

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.12.6

Originally created by @ghost on GitHub (Oct 18, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12690 ### What is the issue? Version 0.12.6 adds web search in the GUI to tool-calling models like Qwen3, but then it returns 102,207 tokens for a simple query, the same query using gpt-oss:20b-cloud web search returns 768 tokens. My test prompt was "Who is the current US president search" and it seems like for Qwen3 it just returns 5 webpages and the entire contents of each, so one of them was the Wikipedia page for US president, which is about 100,000 tokens. For gpt-oss:20b-cloud the formatting of web search results is completely different however, just returning a small 768 token snippet. Is this because qwen3:4b sucks at calling ollama's web search properly? In Ollama's blog post ([https://ollama.com/blog/web-search](url)), they do say it can return thousands of tokens and to set context length to at least 32k, but still, 102,207 tokens is a bit much. Any reason for this? ### Relevant log output ```shell ``` ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.12.6
GiteaMirror added the bug label 2026-05-04 21:40:51 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#70477
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?