Budget — Per Request $1

​

Policy JSON

{
  "cost": {
    "per_request_limit": 1.00,
    "action_on_exceed": "DENY",
    "estimate_before_execution": true
  }
}

policy = {
    "cost": {
        "per_request_limit": 1.00,
        "action_on_exceed": "DENY",
        "estimate_before_execution": True,
    }
}

​

What It Does

• Estimates request cost before sending to the LLM based on input tokens, expected output tokens, and model pricing
• Denies requests where the estimated cost exceeds $1.00
• Pre-execution estimation prevents expensive calls from ever reaching the provider
• Works with all 7 supported providers (OpenAI, Anthropic, Gemini, Bedrock, Azure OpenAI, Cohere, Mistral)

​

Example

{
  "action": "DENY",
  "risk_score": 0.85,
  "reasons": ["per_request_budget_exceeded"],
  "evidence": {
    "per_request_limit": 1.00,
    "estimated_cost": 1.35,
    "model": "gpt-4o",
    "input_tokens": 50000,
    "estimated_output_tokens": 4000,
    "estimation_method": "pre_execution"
  }
}

​

Install

# Install via CLI
tealtiger policy install budget-per-request-1usd

# Or copy the JSON above into your tealtiger.config.json

​

Customize

​

Related

• Budget — Daily $10 — Aggregate daily spending limits
• Model Tier Enforcement — Restrict access to expensive models
• Customer Support Bot Starter — Includes per-request and daily budgets
• Cost Tracking Documentation — Full API reference