CheckPre

AI Chatbot Cost Calculator

What a customer support or assistant chatbot costs to run on Claude, GPT or Gemini per month, with the system prompt cached.

Updated · By Niraj Kumar · 6 sources

Model

Each request

Count text inCount text in
tokens

System prompt, conversation history, documents and tool definitions are all resent on every call.

tokens

The visible answer.

Count tokens from a prompt or JSON payloadShow

Plain text, OpenAI Chat or Responses JSON, or Anthropic Messages JSON. A final assistant message or a response body counts as output. Counted in your browser; nothing is uploaded.

Loading the tokenizer…

Counts use OpenAI's o200k tokenizer, so they are exact for GPT-6 and GPT-5 models, plus a few tokens of message framing. Other providers' tokenizers count the same text differently and do not publish a conversion factor; Anthropic says Claude models from 4.7 on produce about 30% more tokens than earlier Claude models. For an exact count on another model, paste one of its API responses: the usage it reports is shown above.

tokens

Thinking tokens are billed as output. Leave at 0 for models or settings that do not reason.

Volume

PricePrice

150,000 requests over 30 days.

Discounts

%
0%100%

The share of input that repeats exactly at the start of every request, such as a fixed system prompt. The prompt caching calculator prices cache writes and expiry in detail.

Batch APIBatch API
Monthly cost$864.00$0.00576 a request on Claude Sonnet 5
Per 1,000 requests
$5.76
Input
$360.00
Cached input
$54.00
Output
$450.00
Price rank
15th cheapest of 24
Cheapest listed model
Claude Haiku 5.5
$43.20 a month, $820.80 less
Output52%
  • Input42%
  • Cached input6%
  • Output52%
Every model on this workload, per month3,000 tokens in (60% cached), 300 out, 5,000 a day. Hover or focus a row for rates.
  1. 1Claude Haiku 5.5Anthropic$43.20
  2. 2GPT-6 LunaOpenAI$43.20
  3. 3Qwen3.8 FlashAlibaba$88.65
  4. 4GPT-5.6 LunaOpenAI$95.40
  5. 5DeepSeek V4.1 FlashDeepSeek$109.62
  6. 6Qwen3.7 PlusAlibaba$154.80
  7. 7Gemini 3.5 Flash-LiteGoogle$174.60
  8. 8Gemini 3.8 FlashGoogle$324.00
  9. 9Grok 4.3xAI$391.50
  10. 10DeepSeek V4 ProDeepSeek$427.68
  11. 11Claude Haiku 4.5Anthropic$432.00
  12. 12Grok 4.7xAI$765.00
  13. 13Claude Sonnet 5.5Anthropic$837.00
  14. 14GPT-6.1 SolOpenAI$837.00
  15. 15Claude Sonnet 5Anthropic$864.00
  16. 16GPT-6 SolOpenAI$864.00
  17. 17GPT-5.6 TerraOpenAI$954.00
  18. 18Gemini 3.1 ProGoogle$954.00
  19. 19Qwen3.8 MaxAlibaba$1,170
  20. 20Claude Opus 5.5Anthropic$1,674
  21. 21GPT-5.6 SolOpenAI$1,728
  22. 22Claude Opus 5Anthropic$2,160
  23. 23Claude Fable 5.1Anthropic$4,118
  24. 24GPT-6 AstraOpenAI$4,320

How the AI Chatbot Cost Calculator works

cost = (fresh input × input rate + cached input × cache rate + (output + reasoning) × output rate) ÷ 1,000,000

This page uses the same method as the LLM API Cost Calculator, which explains each term and assumption in full.

Worked example

5,000 replies a day. Each sends a 2,000-token system prompt and help-centre extract plus about 1,000 tokens of conversation, and gets a 300-token answer. The fixed system prompt is cached, about 60% of the input. Priced on Claude Sonnet 5.

  1. 1Each request: 3,000 input tokens (60% cached), 300 output
  2. 2Input costs $0.00276 and output $0.003, so $0.00576 a request
  3. 3× 150,000 requests over 30 days = $864.00

$864.00 a month on Claude Sonnet 5. The cheapest listed model, Claude Haiku 5.5, would cost $43.20.

Frequently asked questions

Why does a chatbot's cost grow through a conversation?

The API is stateless, so every reply resends the whole conversation so far as input. The tenth message of a chat carries the nine before it. Average the input over a typical conversation rather than using the first message, and cache the fixed system prompt so the part that repeats most is also the cheapest.

What does one support conversation cost?

Multiply the per-request cost by the replies in a typical conversation. At this page's defaults, a six-reply conversation on a mid-priced model costs a few cents, well below the cost of a human-handled ticket, which is why the model choice matters less than keeping history and system prompts trim.

Sources

Related calculators

More in AI, Cloud & Developer