🎯 Problem Statement
LM Studio currently exposes the OpenAI‑compatible /v1/chat/completions API, but it does not allow callers to request a chain of reasoning steps from the openai/gpt‑oss‑20b model.
For use cases that require transparent explanations (e.g., educational tools, debugging, safety audits), a new optional field – reasoning.effort – would be invaluable.
🔍 Desired Feature
Endpoint: POST /v1/chat/completions
Model: openai/gpt‑oss‑20b
| Field |
Type |
Required? |
Description |
reasoning.effort |
string ("low", "medium", "high" ) |
Optional |
Controls the depth of internal reasoning steps that the model should expose before delivering the final answer. |
When reasoning.effort is present, LM Studio should return a response that includes:
- A textual chain of reasoning (e.g., numbered steps).
- The final answer.
Example Request
POST https://localhost:1234/v1/chat/completions
Content-Type: application/json
{
"model": "openai/gpt‑oss‑20b",
"messages": [
{"role":"user","content":"Explain why the sky is blue."}
],
"reasoning": {
"effort": "high"
}
}Example Response
{
"id": "chatcmpl-01",
"object": "chat.completion",
"created": 1711036800,
"model": "openai/gpt‑oss‑20b",
"choices": [
{
"index": 0,
"message": {
"role":"assistant",
"content":"1️⃣ Light from the sun enters Earth’s atmosphere.\n2️⃣ Short‑wavelength blue light is scattered by air molecules.\n3️⃣ Scattered blue light reaches our eyes from all directions.\n4️⃣ Result: the sky looks blue.",
"reasoning": {
"effort": "high",
"summary": "Light from the sun enters Earth’s atmosphere.\nShort‑wavelength blue light is scattered by air molecules.\nScattered blue light reaches our eyes from all directions.\nResult: the sky looks blue."
},
"final_answer": "The sky appears blue because ..."
}
},
"finish_reason":"stop"
}
],
"usage": {
"prompt_tokens": 12,
"completion_tokens": 48,
"total_tokens": 60
}
}🚧 Why It Matters
- Flexibility: Allows developers to choose between quick answers (
low) or in‑depth explanations (high).
- Cost Management: Users can control token usage by selecting a lower effort level when budgets are tight.
🎯 Problem Statement
LM Studio currently exposes the OpenAI‑compatible
/v1/chat/completionsAPI, but it does not allow callers to request a chain of reasoning steps from theopenai/gpt‑oss‑20bmodel.For use cases that require transparent explanations (e.g., educational tools, debugging, safety audits), a new optional field –
reasoning.effort– would be invaluable.🔍 Desired Feature
Endpoint:
POST /v1/chat/completionsModel:
openai/gpt‑oss‑20breasoning.effort"low","medium","high")When
reasoning.effortis present, LM Studio should return a response that includes:Example Request
Example Response
{ "id": "chatcmpl-01", "object": "chat.completion", "created": 1711036800, "model": "openai/gpt‑oss‑20b", "choices": [ { "index": 0, "message": { "role":"assistant", "content":"1️⃣ Light from the sun enters Earth’s atmosphere.\n2️⃣ Short‑wavelength blue light is scattered by air molecules.\n3️⃣ Scattered blue light reaches our eyes from all directions.\n4️⃣ Result: the sky looks blue.", "reasoning": { "effort": "high", "summary": "Light from the sun enters Earth’s atmosphere.\nShort‑wavelength blue light is scattered by air molecules.\nScattered blue light reaches our eyes from all directions.\nResult: the sky looks blue." }, "final_answer": "The sky appears blue because ..." } }, "finish_reason":"stop" } ], "usage": { "prompt_tokens": 12, "completion_tokens": 48, "total_tokens": 60 } }🚧 Why It Matters
low) or in‑depth explanations (high).