# Can reasoning\_effort parameter not ne used in vllm implementation via python?

**URL:** https://discuss.vllm.ai/t/can-reasoning-effort-parameter-not-ne-used-in-vllm-implementation-via-python/2176
**Category:** Features
**Created:** [January 2, 2026, 10:20am UTC](https://discuss.vllm.ai/t/can-reasoning-effort-parameter-not-ne-used-in-vllm-implementation-via-python/2176 "2026-01-02T10:20:21Z")
**Posts on this page:** 2
**Page:** 1

<div class="post-metadata">

### Author: ![JananiShruthi](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.vllm.ai/jananishruthi/32/1042_2.png) [@JananiShruthi](https://discuss.vllm.ai/u/JananiShruthi)
#### Post date: [January 2, 2026, 10:20am UTC](https://discuss.vllm.ai/t/can-reasoning-effort-parameter-not-ne-used-in-vllm-implementation-via-python/2176/1 "2026-01-02T10:20:21Z")

</div>

I am currently using vllm to infer from the model gpt-oss-20b. I want to keep the model’s reasoning ability to low. I read that vllm support a parameter called reasoning\_effort. Can I use that parameter if I am using vllm via python code and not server?

---

<div class="post-metadata">

### Author: ![RunLLM](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.vllm.ai/runllm/32/233_2.png) [@RunLLM](https://discuss.vllm.ai/u/RunLLM)
#### Post date: [January 2, 2026, 10:20am UTC](https://discuss.vllm.ai/t/can-reasoning-effort-parameter-not-ne-used-in-vllm-implementation-via-python/2176/2 "2026-01-02T10:20:39Z")

</div>

No, you cannot use the reasoning\_effort parameter when running vLLM via the Python LLM class for offline inference; it is only supported through the OpenAI-compatible server endpoints (e.g., /v1/chat/completions or /v1/responses). The reasoning\_effort parameter is not available in the SamplingParams or LLM.generate API for direct Python usage—only in HTTP API requests to the server, as confirmed in multiple issues and documentation entries ([source1](https://github.com/vllm-project/vllm/issues/22359), [source2](https://github.com/vllm-project/vllm/issues/24292), [source3](https://github.com/vllm-project/vllm/issues/22809)).

If you need to control reasoning\_effort, you must run vLLM as a server and send requests via the OpenAI-compatible API. Would you like more detail or example code for using the server API?

Sources:

- [https://github.com/vllm-project/vllm/issues/22359](https://github.com/vllm-project/vllm/issues/22359)
- [https://github.com/vllm-project/vllm/issues/24292](https://github.com/vllm-project/vllm/issues/24292)
- [https://github.com/vllm-project/vllm/issues/22809](https://github.com/vllm-project/vllm/issues/22809)

Hint: Mention @RunLLM in the post for followups.