GPT-6 now calls tools in Chat Completions: use it in Cline and Cherry Studio

Clients that use Chat Completions, such as Cline and Cherry Studio, used to get a 400 as soon as GPT-6 tried to call a tool, so you had to switch models or move to Responses. The same request now returns tool calls.

Published: November 4, 2026
Last verified: November 4, 2026
Status: Active

Key facts

Item	Details
Endpoint	`POST /v1/chat/completions`; Responses (`POST /v1/responses`) works as before
Token	`default`, `vip`, or `svip` group, with Usage first or Usage-based billing
Reasoning effort	Send `reasoning_effort` together with `tools`; accepted values are unchanged
`max_tokens`	Accepted; no need to switch to `max_completion_tokens`
Price	Unchanged; billed per token at each model’s rate, listed in the model catalog

Models covered:

GPT-6: gpt-6-astra, gpt-6-sol, gpt-6-luna
GPT-6.1: gpt-6.1-sol, gpt-6.1-terra
GPT-5.6: gpt-5.6-sol, gpt-5.6-luna, gpt-5.6-terra

What this unlocks

Use case	Before	Now
Cline and other coding agents on OpenAI Compatible	GPT-6 failed; you had to pick another model	Pick `gpt-6.1-sol`
Cherry Studio with MCP tools on	Error; you had to switch models	Use `gpt-6-sol`
Other agents and desktop clients that speak Chat Completions	Every tool call returned 400	Tool calls come back normally
Your own agent built on the OpenAI SDK	Rewrite it for Responses	Keep `chat.completions.create`

How to use it

Add tools to your existing Chat Completions request and set reasoning effort as usual. Install the SDK first (python -m pip install openai) and set the LAOZHANG_API_KEY environment variable:

import os

from openai import OpenAI

client = OpenAI(api_key=os.environ["LAOZHANG_API_KEY"], base_url="https://api.laozhang.ai/v1")

response = client.chat.completions.create(
    model="gpt-6.1-sol",
    reasoning_effort="high",
    messages=[{"role": "user", "content": "What's the weather in Paris right now?"}],
    tools=[{
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get the current weather for a city",
            "parameters": {
                "type": "object",
                "properties": {"city": {"type": "string"}},
                "required": ["city"],
            },
        },
    }],
)

for call in response.choices[0].message.tool_calls:
    print(call.function.name, call.function.arguments)

The script prints the function the model wants to call and its arguments, for example get_weather {"city":"Paris"}. Run the function, send the result back as a tool message, and the model writes the final answer. The full two-turn example is in function calling.

What hasn’t changed

Reasoning effort values: gpt-6-astra and gpt-6.1-sol don’t accept none.
max reasoning effort: in Chat Completions, gpt-6.1-sol goes up to xhigh; use Responses for max.
Sampling parameters: unless reasoning effort is none, leave out temperature, top_p, and similar parameters. See the GPT-6 launch announcement.
Web search: still available only through Responses; see web search.

Who is affected

You send tools to GPT-6 or GPT-5.6 through Chat Completions: requests that used to fail now work. No changes needed.
You moved to Responses for tool calling: it keeps working, and you don’t need to move back.
You set reasoning effort to none so tools would work: that still works. To let the model reason, set low or higher.
Not affected: other models and all Responses requests.

FAQ

Is “Function tools with reasoning_effort are not supported” gone?

Not since November 4, 2026. The error comes from OpenAI’s rules for Chat Completions, and LaoZhang API now handles these requests on its side. If you still see it:

Check that the base URL is https://api.laozhang.ai/v1, not OpenAI’s or another provider’s.
Send support the time of the error, the model, and the full error message.

Can GPT-6.1 Sol call tools without the Responses API?

Yes. On LaoZhang API, gpt-6.1-sol calls tools in Chat Completions, so clients and code that only support Chat Completions can use it.

Does pricing change?

No. Billing is still per token at the model’s rate. Tool definitions and function results count as input tokens, the call arguments the model returns count as output tokens, and there is no separate fee for tool calls.

Should I move my Responses code back to Chat Completions?

No need; both work. If you stay on Responses, follow the multi-turn pattern when you send history back.

Function calling: full two-turn examples for Chat Completions and Responses
GPT-6 launch announcement and GPT-6.1 Sol launch announcement: prices, reasoning effort, and how to choose
Setup guides for Cline and Cherry Studio
OpenAI: Function calling, Using GPT-6

​Key facts

​What this unlocks

​How to use it

​What hasn’t changed

​Who is affected

​FAQ

​Is “Function tools with reasoning_effort are not supported” gone?

​Can GPT-6.1 Sol call tools without the Responses API?

​Does pricing change?

​Should I move my Responses code back to Chat Completions?

​Related documentation