Consult LLM MCP

import { CompletionUsage } from 'openai/resources.js' import { type SupportedChatModel } from './schema.js' // Model pricing data type ModelPricing = { inputCostPerMillion: number outputCostPerMillion: number } const MODEL_PRICING: Partial<Record<SupportedChatModel, ModelPricing>> = { o3: { inputCostPerMillion: 2.0, outputCostPerMillion: 8.0, }, 'gemini-2.5-pro': { inputCostPerMillion: 1.25, outputCostPerMillion: 10.0, }, 'gemini-3-pro-preview': { inputCostPerMillion: 2.0, outputCostPerMillion: 12.0, }, 'deepseek-reasoner': { inputCostPerMillion: 0.55, outputCostPerMillion: 2.19, }, } export function calculateCost( usage: CompletionUsage | undefined, model: SupportedChatModel, ): { inputCost: number; outputCost: number; totalCost: number } { const pricing = MODEL_PRICING[model] if (!pricing) { return { inputCost: 0, outputCost: 0, totalCost: 0 } } const inputTokens = usage?.prompt_tokens || 0 const outputTokens = usage?.completion_tokens || 0 const inputCost = (inputTokens / 1_000_000) * pricing.inputCostPerMillion const outputCost = (outputTokens / 1_000_000) * pricing.outputCostPerMillion const totalCost = inputCost + outputCost return { inputCost, outputCost, totalCost } }