What happened?

I am using local model and my settings.json has:

...
"model": {
    "name": "qwen3.6-27b",
    "generationConfig": {
      "timeout": 60000,
      "enableCacheControl": true,
      "contextWindowSize": 192000,
      "extra_body": {
        "enable_thinking": true,
        "preserve_thinking": true
      },
      "modalities": {
        "image": false
      }
    }
  }
...,

I also set the size in llama.cpp server as same value.

However, in qwen code, the reported size is 1000k, causing the % estimation incorrect at the lower right corner and also auto compression likely won't run because even auto run at 50% of 1000k is 500k which is far greater than the actual context window size.

What did you expect to happen?

Qwen code report correct context window size.

Client information

$ qwen /about
Qwen Code v0.15.6
Model: qwen3.6-27b
Fast Model: not set
Auth: openai
Platform: win32 x64 (10.0.26200)
Node.js: v22.22.2
Session: c4a06941-ac7e-4fba-98b4-e1966465ac1c
Git commit: cf5f447fd

Login information

I am using local server.

Anything else we need to know?

No response