github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 13:19:40 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #15004] Longer thinking time settings? #87330

New Issue
Closed
opened 2026-05-10 05:19:45 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-05-10 05:19:45 -05:00
Owner
Copy Link

Originally created by @SoftologyPro on GitHub (Mar 22, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15004

Is it possible to have longer thinking times beyond High?
Or allow the user to customize longer times.
I would be happy to wait 5x longer to see if the results got better.
High x5 is Extreme
High x10 is 5x Ridiculous
I have been getting some very interesting results with gpt-oss:20b, but allowing longer thinking time may give better results?

Originally created by @SoftologyPro on GitHub (Mar 22, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15004 Is it possible to have longer thinking times beyond High? Or allow the user to customize longer times. I would be happy to wait 5x longer to see if the results got better. High x5 is Extreme High x10 is 5x Ridiculous I have been getting some very interesting results with gpt-oss:20b, but allowing longer thinking time may give better results?
GiteaMirror added the feature request label 2026-05-10 05:19:45 -05:00
GiteaMirror commented 2026-05-10 05:20:44 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 22, 2026):

Since you mention "high" and gpt-oss I assume you are talking about the levels of reasoning it supports - "low", "medium", "high". The amount of thinking the model does is baked into the model, there's no parameter that can be set to increase the thinking output. Instead, you need to use prompt engineering with the model to encourage it to do more thinking. For example, you might add instructions to tell it to think more, or break a prompt into several parts and have the model think about each part, or feed the result of a prompt back into the model and ask it to do more thinking about the previous response.

<!-- gh-comment-id:4106328122 --> @rick-github commented on GitHub (Mar 22, 2026): Since you mention "high" and gpt-oss I assume you are talking about the levels of reasoning it supports - "low", "medium", "high". The amount of thinking the model does is baked into the model, there's no parameter that can be set to increase the thinking output. Instead, you need to use prompt engineering with the model to encourage it to do more thinking. For example, you might add instructions to tell it to think more, or break a prompt into several parts and have the model think about each part, or feed the result of a prompt back into the model and ask it to do more thinking about the previous response.
GiteaMirror commented 2026-05-10 05:20:44 -05:00
Author
Owner
Copy Link

@SoftologyPro commented on GitHub (Mar 23, 2026):

OK, thanks. That explains it all now :)

<!-- gh-comment-id:4109067463 --> @SoftologyPro commented on GitHub (Mar 23, 2026): OK, thanks. That explains it all now :)
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 feature request
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#87330
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?