github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 08:59:13 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12158] Response message after a tool_call with OpenAI chat completions and gpt-oss missing "reasoning" #8083

New Issue
Open
opened 2026-04-12 20:22:46 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-12 20:22:46 -05:00
Owner
Copy Link

Originally created by @pamelafox on GitHub (Sep 2, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12158

Originally assigned to: @ParthSareen on GitHub.

What is the issue?

When we call gpt-oss with the OpenAI chat completions API (/v1/ endpoint), the response contains a choices list, and the first choice has a message, and that message has a reasoning field.

However, when I then execute a tool call and send that longer thread to the model, the response has no reasoning field.

Code to replicate:

import json
import os

import azure.identity
import openai
from dotenv import load_dotenv

load_dotenv(override=True)

client = openai.OpenAI(base_url=os.environ["OLLAMA_ENDPOINT"], api_key="nokeyneeded")
MODEL_NAME = "gpt-oss:20b"


def lookup_weather(city_name=None, zip_code=None):
    """Lookup the weather for a given city name or zip code."""
    return {
        "city_name": city_name,
        "zip_code": zip_code,
        "weather": "sunny",
        "temperature": 75,
    }


tools = [
    {
        "type": "function",
        "function": {
            "name": "lookup_weather",
            "description": "Lookup the weather for a given city name or zip code.",
            "parameters": {
                "type": "object",
                "properties": {
                    "city_name": {
                        "type": "string",
                        "description": "The city name",
                    },
                    "zip_code": {
                        "type": "string",
                        "description": "The zip code",
                    },
                },
                "strict": True,
                "additionalProperties": False,
            },
        },
    }
]

messages = [
    {"role": "system", "content": "You are a weather chatbot."},
    {"role": "user", "content": "is it sunny in berkeley CA?"},
]
response = client.chat.completions.create(
    model=MODEL_NAME,
    messages=messages,
    tools=tools,
    tool_choice="auto",
    reasoning_effort="medium"
)

print(f"Response from {MODEL_NAME} on {API_HOST}: \n")

# Now actually call the function as indicated

print("Thinking...")
print (response.choices[0].message.reasoning)
print("Done thinking.")
if response.choices[0].message.tool_calls:
    tool_call = response.choices[0].message.tool_calls[0]
    function_name = tool_call.function.name
    arguments = json.loads(tool_call.function.arguments)

    if function_name == "lookup_weather":
        messages.append(response.choices[0].message)
        result = lookup_weather(**arguments)
        messages.append({"role": "tool", "tool_call_id": tool_call.id, "content": str(result)})
        print(messages)
        response = client.chat.completions.create(model=MODEL_NAME, messages=messages, tools=tools, reasoning_effort="medium")
        # Potential bug: The follow-up message has NO reasoning field
        #print("Thinking...")
        #print (response.choices[0].message.reasoning)
        #print("Done thinking.")
        print(response.choices[0].message.content)

Relevant log output

OS

Docker, Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.11.8

Originally created by @pamelafox on GitHub (Sep 2, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12158 Originally assigned to: @ParthSareen on GitHub. ### What is the issue? When we call gpt-oss with the OpenAI chat completions API (/v1/ endpoint), the response contains a choices list, and the first choice has a message, and that message has a reasoning field. However, when I then execute a tool call and send that longer thread to the model, the response has *no* reasoning field. Code to replicate: ``` import json import os import azure.identity import openai from dotenv import load_dotenv load_dotenv(override=True) client = openai.OpenAI(base_url=os.environ["OLLAMA_ENDPOINT"], api_key="nokeyneeded") MODEL_NAME = "gpt-oss:20b" def lookup_weather(city_name=None, zip_code=None): """Lookup the weather for a given city name or zip code.""" return { "city_name": city_name, "zip_code": zip_code, "weather": "sunny", "temperature": 75, } tools = [ { "type": "function", "function": { "name": "lookup_weather", "description": "Lookup the weather for a given city name or zip code.", "parameters": { "type": "object", "properties": { "city_name": { "type": "string", "description": "The city name", }, "zip_code": { "type": "string", "description": "The zip code", }, }, "strict": True, "additionalProperties": False, }, }, } ] messages = [ {"role": "system", "content": "You are a weather chatbot."}, {"role": "user", "content": "is it sunny in berkeley CA?"}, ] response = client.chat.completions.create( model=MODEL_NAME, messages=messages, tools=tools, tool_choice="auto", reasoning_effort="medium" ) print(f"Response from {MODEL_NAME} on {API_HOST}: \n") # Now actually call the function as indicated print("Thinking...") print (response.choices[0].message.reasoning) print("Done thinking.") if response.choices[0].message.tool_calls: tool_call = response.choices[0].message.tool_calls[0] function_name = tool_call.function.name arguments = json.loads(tool_call.function.arguments) if function_name == "lookup_weather": messages.append(response.choices[0].message) result = lookup_weather(**arguments) messages.append({"role": "tool", "tool_call_id": tool_call.id, "content": str(result)}) print(messages) response = client.chat.completions.create(model=MODEL_NAME, messages=messages, tools=tools, reasoning_effort="medium") # Potential bug: The follow-up message has NO reasoning field #print("Thinking...") #print (response.choices[0].message.reasoning) #print("Done thinking.") print(response.choices[0].message.content) ``` ### Relevant log output ```shell ``` ### OS Docker, Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.11.8
GiteaMirror added the bug label 2026-04-12 20:22:46 -05:00
GiteaMirror commented 2026-04-12 20:22:48 -05:00
Author
Owner
Copy Link

@lefoulkrod commented on GitHub (Sep 6, 2025):

Possibly related to https://github.com/ollama/ollama/issues/12203

<!-- gh-comment-id:3262164683 --> @lefoulkrod commented on GitHub (Sep 6, 2025): Possibly related to https://github.com/ollama/ollama/issues/12203
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#8083
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?