github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #10928] ollama 0.9.0 can support server level think mode #7191

New Issue
Open
opened 2026-04-12 19:11:22 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 19:11:22 -05:00
Owner
Copy Link

Originally created by @hlstudio on GitHub (May 31, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/10928

The release of ollama 0.9.0 has brought improvements to the thinking mode, which can be enabled or disabled via CLI or API.
Is there any way to keep a specific model in non-thinking mode by default when using it through the serve interface? I'd rather not manually adjust those application settings.
such as
1、Add option (--nothink) for 'ollama serve' that cannot be overridden
2、Allow environment-variable: OLLAMA_THINK

or anyway like hack the model template file ?

Originally created by @hlstudio on GitHub (May 31, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/10928 The release of ollama 0.9.0 has brought improvements to the thinking mode, which can be enabled or disabled via CLI or API. Is there any way to keep a specific model in non-thinking mode by default when using it through the serve interface? I'd rather not manually adjust those application settings. such as 1、Add option (--nothink) for 'ollama serve' that cannot be overridden 2、Allow environment-variable: OLLAMA_THINK or anyway like hack the model template file ?
GiteaMirror added the feature request label 2026-04-12 19:11:22 -05:00
GiteaMirror commented 2026-04-12 19:11:23 -05:00
Author
Owner
Copy Link

@shakti-bhakta commented on GitHub (May 31, 2025):

I also think this would be helpful. I am using Chatbox as the chat interface.
I cannot find any way to disable thinking since the /set nothink command does not work and I also cannot modify the json request sent.

It seem like this suggestion would solve my issue as well.

<!-- gh-comment-id:2925600597 --> @shakti-bhakta commented on GitHub (May 31, 2025): I also think this would be helpful. I am using Chatbox as the chat interface. I cannot find any way to disable thinking since the `/set nothink` command does not work and I also cannot modify the json request sent. It seem like this suggestion would solve my issue as well.
GiteaMirror commented 2026-04-12 19:11:24 -05:00
Author
Owner
Copy Link

@webdev23 commented on GitHub (Jun 3, 2025):

Side note, do not break the API.

Devs added that "thinking" capabilities, ok, great.

But:

In Ollama's API, a model's thinking is now returned as a separate thinking field for easy parsing

Thanks but that is a deep change in the parsing response, so peoples making their own UI and tools have to rewrite response handling.

Please add capabilities but do not break existing previous API calls and responses, so we could implement the newest capabilities changes without the mess.

<!-- gh-comment-id:2936569987 --> @webdev23 commented on GitHub (Jun 3, 2025): Side note, do not break the API. Devs added that "thinking" capabilities, ok, great. But: > In Ollama's API, a model's thinking is now returned as a separate thinking field for easy parsing Thanks but that is a deep change in the parsing response, so peoples making their own UI and tools have to rewrite response handling. Please add capabilities but **do not break existing previous API calls and responses**, so we could implement the newest capabilities changes without the mess.
GiteaMirror commented 2026-04-12 19:11:25 -05:00
Author
Owner
Copy Link

@bakman2 commented on GitHub (Jun 4, 2025):

Thanks but that is a deep change in the parsing response, so peoples making their own UI and tools have to rewrite response handling.

If you do not specify think:true/false in the request, the response will be as previously, ie. should be non-breaking.

<!-- gh-comment-id:2938512486 --> @bakman2 commented on GitHub (Jun 4, 2025): > Thanks but that is a deep change in the parsing response, so peoples making their own UI and tools have to rewrite response handling. If you do not specify `think:true/false` in the request, the response will be as previously, ie. should be non-breaking.
GiteaMirror commented 2026-04-12 19:11:25 -05:00
Author
Owner
Copy Link

@hlstudio commented on GitHub (Jun 12, 2025):

try this modelfile，add \n<think>\n\n</think>\n\n after <｜Assistant｜>

# Modelfile generated by "ollama show"
# To build a new Modelfile based on this, replace FROM with:
FROM deepseek-r1:8b

TEMPLATE """{{- if .System }}{{ .System }}{{ end }}
{{- range $i, $_ := .Messages }}
{{- $last := eq (len (slice $.Messages $i)) 1}}
{{- if eq .Role "user" }}<｜User｜>{{ .Content }}
{{- else if eq .Role "assistant" }}<｜Assistant｜>\n<think>\n\n</think>\n\n
  {{- if and $.IsThinkSet (and $last .Thinking) -}}
<think>
{{ .Thinking }}
</think>
{{- end }}{{ .Content }}{{- if not $last }}<｜end▁of▁sentence｜>{{- end }}
{{- end }}
{{- if and $last (ne .Role "assistant") }}<｜Assistant｜>\n<think>\n\n</think>\n\n
{{- if and $.IsThinkSet (not $.Think) -}}
<think>

</think>

{{ end }}
{{- end -}}
{{- end }}"""
PARAMETER top_p 0.95
PARAMETER stop <｜begin▁of▁sentence｜>
PARAMETER stop <｜end▁of▁sentence｜>
PARAMETER stop <｜User｜>
PARAMETER stop <｜Assistant｜>
PARAMETER temperature 0.6
<!-- gh-comment-id:2964762690 --> @hlstudio commented on GitHub (Jun 12, 2025): try this modelfile，add `\n<think>\n\n</think>\n\n ` after `<｜Assistant｜>` ``` # Modelfile generated by "ollama show" # To build a new Modelfile based on this, replace FROM with: FROM deepseek-r1:8b TEMPLATE """{{- if .System }}{{ .System }}{{ end }} {{- range $i, $_ := .Messages }} {{- $last := eq (len (slice $.Messages $i)) 1}} {{- if eq .Role "user" }}<｜User｜>{{ .Content }} {{- else if eq .Role "assistant" }}<｜Assistant｜>\n<think>\n\n</think>\n\n {{- if and $.IsThinkSet (and $last .Thinking) -}} <think> {{ .Thinking }} </think> {{- end }}{{ .Content }}{{- if not $last }}<｜end▁of▁sentence｜>{{- end }} {{- end }} {{- if and $last (ne .Role "assistant") }}<｜Assistant｜>\n<think>\n\n</think>\n\n {{- if and $.IsThinkSet (not $.Think) -}} <think> </think> {{ end }} {{- end -}} {{- end }}""" PARAMETER top_p 0.95 PARAMETER stop <｜begin▁of▁sentence｜> PARAMETER stop <｜end▁of▁sentence｜> PARAMETER stop <｜User｜> PARAMETER stop <｜Assistant｜> PARAMETER temperature 0.6 ```
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#7191
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?