github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-22 13:42:25 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #14972] Add enable_thinking parameter to disable CoT/Reasoning generation #9626

New Issue
Closed
opened 2026-04-12 22:31:40 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 22:31:40 -05:00
Owner
Copy Link

Originally created by @HavenCTO on GitHub (Mar 20, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14972

Requesting a native boolean flag (enable_thinking: false) to explicitly disable Chain-of-Thought (CoT) reasoning for models like Qwen 3.5 and GPT-OSS.

  • Current Issues:
  • Unreliable Workarounds: System prompts ("Do not think") are often ignored.
  • Performance Hit: Models generate 5-10x more tokens on "thinking" than the actual answer, drastically slowing
  • inference (e.g., 80 t/s → 10 t/s).
  • Truncation: Users must set extremely low num_predict limits, cutting off valid answers mid-sentence.
  • Stability: High reasoning settings frequently cause infinite loops in agents.

Proposed Solution:

Add a dedicated parameter to the API and Modelfile to toggle reasoning generation off, ensuring direct, fast, and stable responses without relying on fragile prompt engineering.

Originally created by @HavenCTO on GitHub (Mar 20, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14972 Requesting a native boolean flag (enable_thinking: false) to explicitly disable Chain-of-Thought (CoT) reasoning for models like Qwen 3.5 and GPT-OSS. - Current Issues: - Unreliable Workarounds: System prompts ("Do not think") are often ignored. - Performance Hit: Models generate 5-10x more tokens on "thinking" than the actual answer, drastically slowing - inference (e.g., 80 t/s → 10 t/s). - Truncation: Users must set extremely low num_predict limits, cutting off valid answers mid-sentence. - Stability: High reasoning settings frequently cause infinite loops in agents. _Proposed Solution:_ **Add a dedicated parameter to the API and Modelfile to toggle reasoning generation off, ensuring direct, fast, and stable responses without relying on fragile prompt engineering.**
GiteaMirror added the feature request label 2026-04-12 22:31:40 -05:00
GiteaMirror commented 2026-04-12 22:31:41 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 20, 2026):

Set "think":false in the API request. gpt-oss can't disable thinking, but it will be reduced.

$ curl -s localhost:11434/api/generate -d '{"model":"qwen3.5","prompt":"hello","think":false,"stream":false}' | jq 'del(.context)'
{
  "model": "qwen3.5",
  "created_at": "2026-03-20T08:55:37.015107541Z",
  "response": "Hello! How can I help you today?",
  "done": true,
  "done_reason": "stop",
  "total_duration": 601682609,
  "load_duration": 506548726,
  "prompt_eval_count": 13,
  "prompt_eval_duration": 26263746,
  "eval_count": 10,
  "eval_duration": 65306368
}
<!-- gh-comment-id:4096674820 --> @rick-github commented on GitHub (Mar 20, 2026): Set `"think":false` in the API request. gpt-oss can't disable thinking, but it will be reduced. ```console $ curl -s localhost:11434/api/generate -d '{"model":"qwen3.5","prompt":"hello","think":false,"stream":false}' | jq 'del(.context)' { "model": "qwen3.5", "created_at": "2026-03-20T08:55:37.015107541Z", "response": "Hello! How can I help you today?", "done": true, "done_reason": "stop", "total_duration": 601682609, "load_duration": 506548726, "prompt_eval_count": 13, "prompt_eval_duration": 26263746, "eval_count": 10, "eval_duration": 65306368 } ```
GiteaMirror commented 2026-04-12 22:31:41 -05:00
Author
Owner
Copy Link

@HavenCTO commented on GitHub (Mar 20, 2026):

Thanks, but this request is for server side disablement of reasoning

<!-- gh-comment-id:4098244641 --> @HavenCTO commented on GitHub (Mar 20, 2026): Thanks, but this request is for server side disablement of reasoning
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#9626
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?