[bug] status line not displaying correct model and context usage 0

[dev-seahouse]

Context:
here is my ~/.claudish/config.json profile config

    "default": {
      "name": "default",
      "models": {
        "opus": "zai@zai@glm-5.1",
        "sonnet": "zai@zai@glm-5.1",
        "haiku": "zai@glm-4.7",
        "subagent": "zai@zai@glm-5.1"
      },
      "createdAt": "2026-01-27T15:11:40.474Z",                                                                                                                                                      
      "updatedAt": "2026-03-26T08:44:32.355Z"
    },

it is set to glm-5.1, but claudish shows Model: zai@glm5 instead of zai@glm5.1

also , notice that the context usage in status line is always 100% when using either glm5 or glm5.1

[claudish-bot]

@dev-seahouse Two separate things going on here.

Model name display: The token file writer in token-tracker.ts:writeFile() doesn't actually include model_name in the JSON it writes — so the status line falls back to CLAUDISH_ACTIVE_MODEL_NAME from the env, which is the raw model ID before resolution. Also, your config has zai@zai@glm-5.1 (double zai@ prefix) which will cause weird parsing — try zai@glm-5.1 instead.

Context usage always 0: The Z.AI provider uses the anthropic transport, and the anthropic SSE parser in anthropic-sse.ts extracts tokens from data.usage and data.message.usage fields. If Z.AI's API doesn't return usage in those fields (or omits them entirely), tokens stay at 0. Could you run with --debug and share the log? That'd confirm whether Z.AI is sending usage data at all.

Also worth noting: glm-5.1 isn't in the model catalog (model-catalog.ts) yet — only glm-5 and glm-4.7 are listed — so the context window would fall through to the generic glm- fallback of 131k.

[kenan-xin]

the context window for glm are as follows:

"models": [
  {
    "id": "glm-5",
    "name": "GLM-5",
    "reasoning": true,
    "input": ["text"],
    "cost": {"input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0},
    "contextWindow": 204800,
  },
  {
    "id": "glm-4.7",
    "name": "GLM-4.7",
    "reasoning": true,
    "input": ["text"],
    "cost": {"input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0},
    "contextWindow": 204800,
  },
  {
    "id": "glm-5.1",
    "name": "GLM-5.1",
    "reasoning": true,
    "input": ["text"],
    "cost": {"input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0},
    "contextWindow": 204800,
  }
]

https://docs.z.ai/devpack/using5.1

[claudish-bot]

@kenan-xin Thanks — that confirms all three GLM models (glm-5, glm-4.7, glm-5.1) should have a 204800 context window. Right now model-catalog.ts only has glm-5 and glm-4.7, and the generic glm- fallback uses 131k, so glm-5.1 gets the wrong window size. Adding glm-5.1 to the catalog with contextWindow: 204800 would fix the 100% context usage issue (it's dividing by too small a window).