Token Plan

Quick start

1. Open the Plan console, sign in, confirm your plan is active, and create or copy its dedicated Plan Key. Do not substitute a standard API key.
2. Use https://gw.r9s.ai/v1 as the OpenAI-compatible API base URL, including /v1. For Anthropic-compatible clients, use https://gw.r9s.ai as the base URL, without /v1.
3. Store the Plan Key on the server or in your tool’s secure credential store. Select glm-5.3-plan or glm-5.3-flash-plan.

Use the client base URL when configuring an SDK, not the full request URL. Anthropic-compatible clients append /v1/messages; do not add /v1 to ANTHROPIC_BASE_URL, or the path may become /v1/v1/messages. API requests require your Plan Key; opening an endpoint link in a browser does not authenticate the request.

export R9S_PLAN_BASE_URL="https://gw.r9s.ai/v1"
export R9S_PLAN_KEY="<R9S_PLAN_KEY>"

curl "${R9S_PLAN_BASE_URL%/}/chat/completions" \
  -H "Authorization: Bearer $R9S_PLAN_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.3-plan",
    "messages": [{"role": "user", "content": "Hello from Token Plan."}]
  }'

Use glm-5.3-flash-plan in the same request to select GLM-5.3-Flash. Both models share your plan’s billing-cycle allowance and rolling limits. Check model availability with the same Plan Key, not a different account key.

For tools that read Anthropic environment variables, set the Anthropic-compatible base URL and select the Plan models explicitly:

export ANTHROPIC_BASE_URL="https://gw.r9s.ai"
export ANTHROPIC_AUTH_TOKEN="<R9S_PLAN_KEY>"
export ANTHROPIC_MODEL="glm-5.3-plan"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-5.3-flash-plan"

These variables are client-specific settings, not a guarantee that every Anthropic SDK reads them automatically. Use your client’s documented base URL, authentication, and model settings.

A plan_model_required error means the Plan Key was paired with a non-Plan model. A plan_key_required error means a Plan-only model was paired with another key type. Correct the key/model pair instead of retrying it unchanged.

How included usage is measured

Included GLM-5.3 and GLM-5.3-Flash usage is measured at Token Plan rates: 1x off-peak; 2x from 14:00–18:00 (UTC+8). It is not cash or wallet balance.

The models have different base Token Plan rates. For an identical model and token mix, usage that consumes $1 of allowance off-peak consumes $2 during the peak window. The multiplier changes allowance consumption, not your plan’s fixed purchase price. When a limit is reached, requests pause until the applicable limit resets; there are no overage charges. Confirm the active rates, limits, and purchased terms in the Console.