Saved
AI & LLM tools·Cost Tools

LLM Cost Comparator

Compare AI model API costs side by side. Enter your token usage and see GPT-5, Claude 4, and Gemini 3 prices ranked cheapest to most expensive for your.

Added May 26, 2026 · Updated May 27, 2026

Quick examples

Input

Result

Enter a value for input tokens per request to see your result.

How it works

Compare API pricing for GPT-5, Claude 4, Gemini 3, and more side by side. Enter your average token usage and daily request volume to instantly see all models ranked cheapest to most expensive. Optionally add a second scenario to rank models by combined monthly cost across two workloads.

Step by step

  1. 01Enter the average number of input tokens per request — your prompt, system message, and conversation history combined.
  2. 02Enter the average number of output tokens — the expected response length from the model.
  3. 03Set your daily request volume (API calls per day).
  4. 04Optional: enable Compare a second scenario to model two workloads (e.g. chatbot + RAG) and rank models by combined monthly cost.
  5. 05The comparison table ranks all models from cheapest to most expensive for your workload(s).

Examples

High-volume chatbot: 1,000 req/day, 500 input + 300 output tokens

At 1,000 requests/day with short messages, mini-class models cost 10–20× less than premium options. Switching from a premium to a budget model here saves over $1,000/month.

Inputs

Input tokens per request:
500
Output tokens per request:
300
Requests per day:
1000

Result

Most cost-effective model:
GPT-5.4 mini
Monthly cost (cheapest):
51.75
Note: Prices are from data/model-pricing.json, refreshed on each production build. Always verify at the provider's pricing page before committing budgets. Llama 4 Maverick is open-weights and is excluded — API cost depends entirely on your hosting provider or self-hosting setup. Prompt caching (available on Anthropic and Google) is not included. For apps with a large fixed system prompt, real input costs can be 50–90% lower.

Related guides

Frequently asked questions

Which AI model API is cheapest?

It depends on your token mix and volume. For short-context, high-volume workloads, GPT-5.4 mini and Claude Haiku 4.5 are typically the most cost-effective hosted options. For long-context tasks, compare carefully — output token pricing often dominates total cost.

Why do output tokens cost more than input tokens?

Generating tokens requires significantly more compute than reading them. Models must run the full forward pass for every output token, whereas input tokens are processed in parallel. Output pricing is usually 3–6× higher per million tokens.

Does this tool include prompt caching discounts?

No. Anthropic and Google offer prompt caching that can reduce repeated input costs by 50–90%. If your application has a large fixed system prompt, enable caching and use the AI API Cost Calculator to estimate your effective rate.

How often are prices updated?

Prices are refreshed from public sources on each production build. The displayed prices reflect the last build date. Major model price changes happen every few months — check the provider's pricing page for the latest rates.

How does comparing two scenarios work?

Enable Compare a second scenario and enter token counts and daily volume for a second workload (e.g. a lightweight chatbot plus a heavy RAG pipeline). The table ranks models by combined monthly cost so you can pick one model that minimizes total spend across both use cases.