Input vs output token spend
Output tokens usually cost several times more than input, so answer length is what really drives your bill.
Home / Dev & AI / Claude API Cost Calculator
AI & LLM Cost
Estimate your Anthropic bill fast. Pick a Claude model, enter your request volume and token sizes, and see the monthly cost with a comparison across Fable, Opus, Sonnet and Haiku for 2026.
Estimated monthly cost
$0
Input vs output token spend
Output tokens usually cost several times more than input, so answer length is what really drives your bill.
The same workload priced across AI models
Your monthly token volume run through popular 2026 models, cheapest first.
Anthropic bills the Claude API per token, separately for input (your prompt, system prompt, documents and conversation history) and output (the model's reply, including any thinking it does). The current lineup has four tiers. Claude Haiku 5.5 is the fast, very low-cost option for high-volume work. Claude Sonnet 5.5 is the everyday workhorse for coding, agents and enterprise apps. Claude Opus 5.5 is the default high-intelligence model. Claude Fable 5.1 is Anthropic's most capable widely released model, for the hardest reasoning and long agentic tasks.
This calculator prices your workload on the model you choose and compares it across all four, so you can see what moving up or down a tier would do to your monthly bill.
On Claude Sonnet 5.5, 100,000 requests a month at 1,000 input and 500 output tokens is 100M input tokens and 50M output tokens: 100 × $2 + 50 × $10 = $200 + $500 = about $700 a month. The same workload costs about $35 on Claude Haiku 5.5, $1,400 on Claude Opus 5.5 and $3,500 on Claude Fable 5.1.
| Model | Input / 1M | Output / 1M | Context window |
|---|---|---|---|
| Claude Fable 5.1 | $10.00 | $50.00 | 1M tokens |
| Claude Opus 5.5 | $4.00 | $20.00 | 1M tokens |
| Claude Sonnet 5.5 | $2.00 | $10.00 | 1M tokens |
| Claude Haiku 5.5 | $0.10* | $0.50* | 1M tokens |
*Haiku 5.5 prices apply to prompts up to 100K tokens. Longer prompts are billed at $0.50 input and $2.50 output per million. Anthropic first-party list prices, checked October 2026. Confirm on Anthropic's pricing page before budgeting.
A token is a piece of a word. In English, one token is roughly three-quarters of a word, or about four characters, so 1,000 tokens is around 750 words. Code, non-English text and data such as JSON usually take more tokens per word. Useful reference points for estimating your inputs:
| Content | Approx. tokens |
|---|---|
| A short chat message | 20–100 |
| A one-page email or memo (500 words) | about 650 |
| A detailed system prompt | 1,000–5,000 |
| A 10-page PDF | 5,000–8,000 |
| A 300-page book | about 120,000 |
Remember that a chat app resends the whole conversation on every turn. By the tenth message, a single request can carry thousands of input tokens of history. Agents that call tools also add the tool definitions and every tool result to the input on each step. For an exact count, Anthropic's API has a free token-counting endpoint.
effort setting (low to max) that controls how much thinking and how many output tokens the model spends. Lower effort is often plenty for chat, extraction and classification.A support bot on Sonnet 5.5 sends a 4,000-token system prompt and knowledge snippet with every one of 200,000 monthly requests. Uncached, that's 800M input tokens, or $1,600. With caching, almost all of those tokens are read from the cache at $0.20 per million, about $160, plus a small cost for writing the cache. The saving is over $1,400 a month before counting any change to output.
Per million tokens: Claude Haiku 5.5 $0.10 input / $0.50 output, Sonnet 5.5 $2 / $10, Opus 5.5 $4 / $20, and Fable 5.1 $10 / $50. Enter your usage above for a tailored figure.
Opus 5.5 is a strong default for demanding work, and Sonnet 5.5 suits most everyday coding, agent and enterprise tasks. Use Haiku 5.5 for high-volume, latency-sensitive tasks, and Fable 5.1 when you need the most capable model.
Yes. Cached input is read at a fraction of the normal input price, which is a large saving for chatbots, agents and repeated long documents.
Yes. The Message Batches API processes requests asynchronously within 24 hours at 50% of standard prices.
Generating each output token takes more compute than reading input. On every current Claude model, output tokens cost five times as much as input tokens.
They are Anthropic's first-party list prices, checked October 2026. Prices on Amazon Bedrock and Google Vertex AI are set by those platforms. Confirm current rates on Anthropic's pricing page.