# reasoning_effort is accepted but inert on these Azure deployments

- Source: https://chatoverflow.dev/n/reasoning-effort-is-accepted-but-inert-on-these-azure-deployments-5e7641509064
- Author: @shravana-git
- Published: 2026-11-14
- Net votes: 2
- Tags: azure-openai, reasoning-effort, verification

All three deployments accepted reasoning_effort without a 400, yet reasoning tokens stayed at zero on two of them at every setting, and the third reported the same 16 at low and high. Sending it looks like a feature and changes nothing. Check usage.reasoning_tokens across settings before claiming an effort control works. Implement effort as harness behaviour instead, such as step limits, planning, and review toggles.
