LLM API Cost Calculator (OpenAI, Claude, Gemini & DeepSeek)

The LLM API Cost Calculator helps developers, AI founders, and engineers estimate and compare API costs across top Large Language Models (LLMs) including OpenAI (GPT-4o, GPT-4o-mini), Anthropic (Claude 3.5 Sonnet, Claude 3 Opus), Google (Gemini 1.5 Pro, Gemini 1.5 Flash), and DeepSeek (V3 & R1).

### 📊 Token Cost Comparison Table (Pricing per 1M Tokens in USD)

API token pricing is split into Input Tokens (context/prompt) and Output Tokens (generation/response):

| Model | Input Rate / 1M | Output Rate / 1M | Price Ratio (Out : In) |

|---|---|---|---|

| Claude 3 Opus | $15.00 | $75.00 | 5.0x |

| Claude 3.5 Sonnet | $3.00 | $15.00 | 5.0x |

| OpenAI GPT-4o | $2.50 | $10.00 | 4.0x |

| Gemini 1.5 Pro | $1.25 | $5.00 | 4.0x |

| OpenAI GPT-4o-mini | $0.15 | $0.60 | 4.0x |

| DeepSeek V3 / R1 | $0.14 | $0.28 | 2.0x |

### 💼 Case Study: How ChatFlow Cut API Costs by 48% Using Prompt Caching

ChatFlow is a customer support chatbot that handles 10,000 requests per day. Each request passes a 3,000-word system prompt (containing product guidelines) and generates a 150-word response.

1. Scenario A (Standard API Calls): Using Claude 3.5 Sonnet without caching, ChatFlow consumed 4,000 input tokens ($12.00) and 200 output tokens ($3.00) per request.

- Standard Monthly Cost: ($12.00 + $3.00) × 10,000 × 30 = $4,500/month

2. Scenario B (With Prompt Caching): Since the 3,000-word product guidelines remain constant, ChatFlow enabled Prompt Caching. Input cache hits are charged at a 90% discount ($0.30/1M tokens) on Anthropic.

- Cached Monthly Cost: [($1.20 cached input + $0.30 fresh input) + $3.00 output] × 10,000 × 30 = $1,350/month

*By implementing prompt caching, ChatFlow achieved a 70% cost reduction on input tokens, saving over $3,150 every single month!*

### 🧠 Developer Decision Guide

* High-Reasoning Tasks: Use Claude 3.5 Sonnet or DeepSeek R1 for complex coding, agentic workflows, and multi-step reasoning.

* Conversational/Drafting Tasks: Use GPT-4o-mini or Gemini 1.5 Flash for low-cost, low-latency needs. At $0.15/1M tokens, it is nearly 20x cheaper than flagship models.

* DeepSeek V3 Edge Case: DeepSeek V3 offers flagship intelligence (competing with GPT-4o) at a price ratio of only 2.0x (instead of 4x-5x), making output-heavy tasks extremely cost-efficient.

This calculator estimates monthly LLM API costs through a multi-variable developer model:

1. Enter Average Words per Request: Set the expected word count for your prompt (Input) and model response (Output).

2. Set Daily Request Volume: Input how many API calls your application makes per day.

3. Toggle Caching & Batch Discounts: Apply Prompt Caching (reduces input token price by up to 50%-90%) or Batch API (50% flat discount on all tokens).

4. Compare Side-by-Side: Review the estimated monthly bill across all major models to optimize your infrastructure budget.