LLM API Cost Calculator

This calculator uses the standard 4-characters-per-token approximation used by most LLM providers. For English prose, accuracy is typically ±10%. Code, JSON, and non-English text are more token-dense and may have 15–20% more tokens than estimated.

Output tokens require the model to generate each token sequentially, which is computationally more expensive than reading input tokens in parallel. Output costs are typically 3–10× higher than input costs.

For most customer support use cases with short messages, GPT-4o mini ($0.15/$0.60 per 1M), Claude Haiku 4.5 ($1.00/$5.00), or Gemini Flash-Lite ($0.075/$0.30) offer the best cost-to-quality ratio. Use the table above sorted by Total Cost to compare.

Prices are updated manually based on official provider pricing pages as of March 2026. LLM pricing changes frequently — always verify against the provider's official pricing page before making budget decisions.

Key strategies include: (1) Use a smaller model like GPT-4o mini or Gemini Flash for simpler tasks, (2) Enable Batch APIs for 50% off when latency is not critical, (3) Enable prompt caching to save up to 90% on repeated system prompts, (4) Compress your prompts, (5) Use structured output to reduce verbosity.

AWS Bedrock lets you run foundation models (Claude, Llama, Mistral, etc.) within AWS's infrastructure. For Claude models, prices typically match Anthropic direct. For Amazon-native models like Nova Micro, Bedrock can be 80–90% cheaper than GPT-4o.

Multiply (input tokens + output tokens) per request × requests per day × 30 × price per 1M tokens. This calculator does that automatically — just paste a sample prompt and set your expected output length and daily volume.

It depends on the model tier. GPT-4o mini ($0.15 input / $0.60 output per 1M tokens) is cheaper than Claude Sonnet 4.5 ($3/$15 per 1M tokens), but Claude Haiku 4.5 ($1/$5) sits in between. Use this calculator with your actual token counts to compare.