compare_costs
Compare pricing between different AI models to optimize token usage and manage API costs effectively.
Instructions
Compare costs between models
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| tokens | Yes | Number of tokens to compare |
Implementation Reference
- src/mcp-server.ts:303-341 (handler)The main handler function that implements the compare_costs tool. It simulates token usage for various models using the TokenTracker, calculates costs, sorts them, and returns a formatted comparison.private compareCosts(args: any) { const { tokens } = args; const models = [ { provider: 'openai' as const, model: 'gpt-3.5-turbo', name: 'GPT-3.5' }, { provider: 'openai' as const, model: 'gpt-4', name: 'GPT-4' }, { provider: 'anthropic' as const, model: 'claude-3-haiku-20240307', name: 'Claude Haiku' }, { provider: 'anthropic' as const, model: 'claude-3-sonnet-20240229', name: 'Claude Sonnet' }, { provider: 'anthropic' as const, model: 'claude-3-opus-20240229', name: 'Claude Opus' }, { provider: 'gemini' as const, model: 'gemini-1.5-flash', name: 'Gemini Flash' }, { provider: 'gemini' as const, model: 'gemini-1.5-pro', name: 'Gemini Pro' } ]; const comparison = models.map(({ provider, model, name }) => { const trackingId = this.tracker.startTracking('_compare'); this.tracker.endTracking(trackingId, { provider, model, inputTokens: Math.floor(tokens / 2), outputTokens: Math.ceil(tokens / 2), totalTokens: tokens }); const usage = this.tracker.getUserUsage('_compare'); const cost = usage?.usageByModel[`${provider}/${model}`]?.cost || 0; this.tracker.clearUserUsage('_compare'); return { name, cost }; }).sort((a, b) => a.cost - b.cost); let result = `💰 Cost comparison for ${tokens} tokens:\n\n`; comparison.forEach((item, i) => { const emoji = i === 0 ? '🏆' : i === 1 ? '🥈' : i === 2 ? '🥉' : ' '; result += `${emoji} ${item.name}: ${formatCost(item.cost)}\n`; }); return { content: [{ type: 'text', text: result }] }; }
- src/mcp-server.ts:116-125 (schema)Input schema for the compare_costs tool, requiring a 'tokens' number parameter.inputSchema: { type: 'object', properties: { tokens: { type: 'number', description: 'Number of tokens to compare' } }, required: ['tokens'] }
- src/mcp-server.ts:113-126 (registration)Tool registration in the ListTools response, including name, description, and input schema.{ name: 'compare_costs', description: 'Compare costs between models', inputSchema: { type: 'object', properties: { tokens: { type: 'number', description: 'Number of tokens to compare' } }, required: ['tokens'] } },
- src/mcp-server.ts:166-167 (registration)Dispatch case in the CallToolRequest handler that routes to the compareCosts method.case 'compare_costs': return this.compareCosts(request.params.arguments);