Vikey Auto picks the right model for each message

1. 1

   A tier is picked for each new message

   When you send a new message, Vikey picks a light, medium or heavy model based on what the message asks for.

2. 2

   The same model until the task is done

   While the agent works through its tool calls, the model does not change. The next pick happens on your next message.

3. 3

   Helper requests go to the light model

   Small requests from the agent, such as writing a session title, go straight to the light model with no selection fee.

4. 4

   Old history is trimmed

   Tool output and history that no longer matter are trimmed before resending, so the next turn's prompt can shrink by up to about half.

claude --model vikey/agent-gpt

These numbers come from our own tests with real coding agents. Savings are largest in long sessions and when the topic changes. Short sessions save little.

Rp10per new message

The model selection fee, charged only when the request succeeds. Tokens are billed at the price of the model that was picked, at pay-as-you-go prices. There is no monthly fee.

export ANTHROPIC_BASE_URL=https://api.vikey.ai
export ANTHROPIC_AUTH_TOKEN=YOUR_VIKEY_API_KEY
export ANTHROPIC_MODEL=vikey/agent-gpt
export ANTHROPIC_DEFAULT_HAIKU_MODEL=openai/gpt-6-luna
claude

# First run: export VIKEY_API_KEY=YOUR_VIKEY_API_KEY
model = "vikey/agent-gpt"
model_provider = "vikey"

[model_providers.vikey]
name = "Vikey"
base_url = "https://api.vikey.ai/v1"
env_key = "VIKEY_API_KEY"
wire_api = "responses"

// First run: export VIKEY_API_KEY=YOUR_VIKEY_API_KEY
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "vikey": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Vikey",
      "options": {
        "baseURL": "https://api.vikey.ai/v1",
        "apiKey": "{env:VIKEY_API_KEY}"
      },
      "models": {
        "vikey/agent-gpt": {},
        "openai/gpt-6-luna": {}
      }
    }
  },
  "model": "vikey/vikey/agent-gpt",
  "small_model": "vikey/openai/gpt-6-luna"
}

curl https://api.vikey.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_VIKEY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "vikey/agent-gpt", "messages": [{"role": "user", "content": "Halo"}]}'

Cline, Roo Code, Kilo Code

Choose the OpenAI Compatible provider, then enter the base URL, API key and model id.

OpenClaw

Add Vikey as a custom provider with the base URL and API key.

Hermes

Choose a custom OpenAI-compatible endpoint and enter the base URL and API key.

• You work with Claude Code, Codex, opencode or another coding agent.
• Your sessions are long and the topic changes often.
• You would rather not choose a model for each message.
• Short sessions save little. If you always need one specific model, select that model directly.

• Pay as you goPay per token for the model you choose, at prices far below the vendor's official price.See prices for all models
• Token PlanA monthly plan with one model id, vikey/plan. Vikey handles the model, so you never work out token prices.See Token Plan details