Z.AI Cost Calculation for Coding Plan Shows Wrong Value

[orcunilbeyli]

Description

In the Z.AI Coding Plan, usage is limited per prompt within N hours. However, due to the provider.json settings, cost is calculated as if it were per million input/output tokens. Also, the API endpoint is correctly configured automatically by crush.

Expected result is not showing any cost at all while using z.ai coding plan.

"name": "Z.AI",
    "id": "zai",
    "api_key": "$ZAI_API_KEY",
    "api_endpoint": "https://api.z.ai/api/coding/paas/v4",
    "type": "openai",
    "default_large_model_id": "glm-4.5",
    "default_small_model_id": "glm-4.5-air",
    "models": [
      {
        "id": "glm-4.5",
        "name": "GLM-4.5",
        "cost_per_1m_in": 0.6,
        "cost_per_1m_out": 2.2,
        "cost_per_1m_in_cached": 0.11,
        "cost_per_1m_out_cached": 0,
        "context_window": 131072,
        "default_max_tokens": 98304,
        "can_reason": true,
        "has_reasoning_efforts": false,
        "supports_attachments": false
      }

Version

0.9.2

Environment

macos 26.0