llm-costprice

Calculate the cost of any LLM API call in one line.

Zero dependencies. TypeScript. Supports OpenAI, Anthropic, Google, Mistral, Cohere, Meta (via API providers), and more.

Install

npm install llm-costprice

Quick Start

import { calculateCost } from "llm-costprice";

const result = calculateCost({
  model: "gpt-4o",
  inputTokens: 1500,
  outputTokens: 500,
});

console.log(result.totalCost);   // 0.01125
console.log(result.breakdown);    // { inputCost: 0.00375, outputCost: 0.0075, ... }

API

calculateCost(usage: LLMUsage): CostResult

Primary function. Takes a usage object (model + token/character/audio counts) and returns the cost in USD with a breakdown.

import { calculateCost } from "llm-costprice";

const result = calculateCost({
  model: "claude-sonnet-4-20250514",
  inputTokens: 1000,
  outputTokens: 500,
  cachedInputTokens: 200,  // optional, for prompt caching
});
// result.totalCost, result.breakdown, result.model, result.currency

estimateCost(usage: LLMUsage): CostResult

Alias for calculateCost. Use whichever name fits your codebase.

getCost(model: string, inputTokens: number, outputTokens: number): number

Returns only the total cost as a number.

import { getCost } from "llm-costprice";

const cost = getCost("gpt-4o", 1000, 500);

getModel(modelId: string): ModelPricing | undefined

Look up a model's pricing info. Returns undefined if not found (does not throw).

import { getModel } from "llm-costprice";

const model = getModel("gpt-4o");
console.log(model?.inputCostPer1MTokens);

listModels(provider?: string): ModelPricing[]

List all models, optionally filtered by provider.

import { listModels } from "llm-costprice";

const all = listModels();
const anthropic = listModels("anthropic");

listProviders(): string[]

Returns a sorted array of all provider names.

addModel(pricing: ModelPricing): void

Register a custom model at runtime. Overwrites if modelId already exists.

removeModel(modelId: string): boolean

Remove a model from the registry. Returns true if it existed.

formatCost(usd: number, options?: FormatOptions): string

Format a cost number into a human-readable string.

import { formatCost } from "llm-costprice";

formatCost(0.00375);                      // "$0.003750"
formatCost(0.00375, { decimals: 4 });     // "$0.0037"
formatCost(1.5, { symbol: false });       // "1.500000"

Supported Models

Pricing is maintained for all major providers. Use listModels() or listModels("openai") for the full list.

| Provider | Examples | |-----------|----------| | OpenAI | gpt-4o, gpt-4o-mini, gpt-4-turbo, gpt-3.5-turbo, o1, o3, o4-mini | | Anthropic | claude-opus-4, claude-sonnet-4, claude-haiku-4-5, claude-3-5-sonnet | | Google | gemini-2.5-pro, gemini-2.5-flash, gemini-2.0-flash, gemini-1.5-pro | | Mistral | mistral-large-latest, mistral-small-latest, codestral-latest | | Cohere | command-r-plus, command-r, command-light | | Together | llama-3.1-405b, llama-3.1-70b, llama-3.1-8b |

Model IDs are resolved case-insensitively. Many models support aliases (e.g. gpt-4o-2024-08-06) and date-suffix fallback (e.g. gpt-4o-2025-01-15 → gpt-4o).

Custom Models

Register your own pricing (e.g. fine-tuned or private models) at runtime:

import { addModel, calculateCost } from "llm-costprice";

addModel({
  provider: "custom",
  modelId: "my-fine-tune",
  inputCostPer1MTokens: 10,
  outputCostPer1MTokens: 30,
  lastVerified: "2026-01-15",
});

calculateCost({ model: "my-fine-tune", inputTokens: 1000, outputTokens: 500 });

Formatting

Use formatCost for display:

import { formatCost, calculateCost } from "llm-costprice";

const result = calculateCost({ model: "gpt-4o", inputTokens: 1000, outputTokens: 500 });
console.log(formatCost(result.totalCost));

Error Handling

• UnknownModelError — The model ID is not in the registry. The message may include suggestions for similar model names.
• InvalidUsageError — Invalid usage (e.g. negative tokens, cachedInputTokens > inputTokens, or no usage dimensions).

import { calculateCost, UnknownModelError, InvalidUsageError } from "llm-costprice";

try {
  calculateCost({ model: "gpt-4o", inputTokens: -1, outputTokens: 0 });
} catch (e) {
  if (e instanceof InvalidUsageError) {
    console.error("Invalid usage:", e.message);
  }
}

FAQ

• How do I update pricing? Open a PR that updates src/data/models.ts, or call addModel() at runtime. See PRICING_UPDATE.md for contributor guidance.
• Is this accurate? Prices are intended to match provider pricing pages. Each model has a lastVerified date; verify critical numbers yourself.
• Browser support? Yes. No Node-specific APIs; works in browsers and Node 16+.

Contributing

Contributions are welcome. For price updates, see .github/PRICING_UPDATE.md.

License

MIT