Token Counter
count_tokensCount tokens in text across multiple LLM models to check context window usage and compare costs before sending to an LLM.
Instructions
Count tokens for any text across multiple LLM models and get per-model cost estimates. Use before sending text to an LLM to check context window usage or compare costs across models. Supports GPT-4o, GPT-4, GPT-3.5-turbo, Claude 3.5 Sonnet, Claude 3 Opus, and 10+ more.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| text | Yes | The text to count tokens for | |
| models | No | Models to count tokens for |
Implementation Reference
- mcp-server/src/index.ts:405-444 (handler)The "count_tokens" tool is registered here. It uses the helper function `callToolApi` to forward the request to the `token-counter` endpoint of the Agent Toolbelt API.
// ----- Tool: Token Counter ----- server.registerTool( "count_tokens", { title: "Token Counter", description: "Count tokens for any text across multiple LLM models and get per-model cost estimates. " + "Use before sending text to an LLM to check context window usage or compare costs across models. " + "Supports GPT-4o, GPT-4, GPT-3.5-turbo, Claude 3.5 Sonnet, Claude 3 Opus, and 10+ more.", inputSchema: { text: z.string().describe("The text to count tokens for"), models: z .array(z.string()) .default(["gpt-4o", "claude-3-5-sonnet"]) .describe("Models to count tokens for"), }, }, async ({ text, models }) => { const result = await callToolApi("token-counter", { text, models }); const data = result as any; const r = data.result; const lines = [ `**Characters:** ${r.characterCount.toLocaleString()}`, `**Words:** ${r.wordCount.toLocaleString()}`, "", "**Token counts:**", ]; for (const [model, info] of Object.entries(r.results) as any) { const approx = info.approximate ? " (approx)" : ""; const cost = info.estimatedCost ? ` | input ~$${info.estimatedCost.input} / output ~$${info.estimatedCost.output}` : ""; lines.push(` ${model}: **${info.tokens.toLocaleString()} tokens**${approx}${cost}`); } return { content: [{ type: "text" as const, text: lines.join("\n") }] }; } );