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 #8544] API parameter: 'reasoning_effort' (for DeepSeek-R1) #5512

New Issue
Closed
opened 2026-04-12 16:45:48 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 16:45:48 -05:00
Owner
Copy Link

Originally created by @jonathanhecl on GitHub (Jan 23, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8544

When I will be able to use reasoning_effort like on DeepSeek-R1? :)

Doc: https://api-docs.deepseek.com/guides/reasoning_model

Originally created by @jonathanhecl on GitHub (Jan 23, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8544 When I will be able to use `reasoning_effort` like on DeepSeek-R1? :) Doc: https://api-docs.deepseek.com/guides/reasoning_model
GiteaMirror added the feature request label 2026-04-12 16:45:48 -05:00
GiteaMirror commented 2026-04-12 16:45:50 -05:00
Author
Owner
Copy Link

@jonathanhecl commented on GitHub (Jan 27, 2025):

any support? 👍

<!-- gh-comment-id:2615027051 --> @jonathanhecl commented on GitHub (Jan 27, 2025): any support? 👍
GiteaMirror commented 2026-04-12 16:45:52 -05:00
Author
Owner
Copy Link

@Skit5 commented on GitHub (Jan 29, 2025):

Well at least you got my support but I don't know if the reasoning_effort parameter can be used on the local model as it's not available yet even on their API: https://api-docs.deepseek.com/guides/reasoning_model

<!-- gh-comment-id:2621631207 --> @Skit5 commented on GitHub (Jan 29, 2025): Well at least you got my support but I don't know if the `reasoning_effort` parameter can be used on the local model as it's not available yet even on their API: https://api-docs.deepseek.com/guides/reasoning_model
GiteaMirror commented 2026-04-12 16:45:52 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 5, 2026):

deepseek-r1 returns Chain of Thought (CoT) traces as described in https://api-docs.deepseek.com/guides/reasoning_model. reasoning_effort is not a supported parameter either in the deepseek API or the ollama API, although ollama does allow disabling reasoning with the think parameter.

$ ollama run deepseek-r1
>>> hello
Thinking...
Okay, the user just said “hello” – super simple and neutral. 

Hmm, this could be their first interaction with me, or they might be testing the 
waters. Either way, keeping it friendly and open-ended feels right. 

I should match their casual tone but also subtly encourage them to say more. The 
“Hi there!” feels warm but not overly familiar, and the “I'm here to help” sets a 
clear expectation. Adding an emoji (😊) makes it more approachable without being 
unprofessional. 

The user didn't give any context, so I can't assume their needs. Best to keep the 
door wide open for whatever they might want – from deep chat to quick facts. The 
“Let me know” at the end is a gentle nudge to engage further. 

Wonder if they're just saying hello to check if I'm responsive, or if they have 
something specific in mind. Either way, this response covers both scenarios.
...done thinking.

Hi there! 😊  
I'm here to help—just let me know what you need.

>>> /set nothink
Set 'nothink' mode.
>>> hello
Hey there! 😊  
It's nice to hear from you again. How can I assist you today?
<!-- gh-comment-id:3709410839 --> @rick-github commented on GitHub (Jan 5, 2026): deepseek-r1 returns Chain of Thought (CoT) traces as described in https://api-docs.deepseek.com/guides/reasoning_model. `reasoning_effort` is not a supported parameter either in the deepseek API or the ollama API, although ollama does allow disabling reasoning with the `think` parameter. ```console $ ollama run deepseek-r1 >>> hello Thinking... Okay, the user just said “hello” – super simple and neutral. Hmm, this could be their first interaction with me, or they might be testing the waters. Either way, keeping it friendly and open-ended feels right. I should match their casual tone but also subtly encourage them to say more. The “Hi there!” feels warm but not overly familiar, and the “I'm here to help” sets a clear expectation. Adding an emoji (😊) makes it more approachable without being unprofessional. The user didn't give any context, so I can't assume their needs. Best to keep the door wide open for whatever they might want – from deep chat to quick facts. The “Let me know” at the end is a gentle nudge to engage further. Wonder if they're just saying hello to check if I'm responsive, or if they have something specific in mind. Either way, this response covers both scenarios. ...done thinking. Hi there! 😊 I'm here to help—just let me know what you need. >>> /set nothink Set 'nothink' mode. >>> hello Hey there! 😊 It's nice to hear from you again. How can I assist you today? ```
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#5512
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?