Claude API pricing calculator: cost per request, day and month
Estimate what the Claude API costs for your workload: pick a model, enter input and output tokens and request volume, and add prompt caching or the Batch API. Prices from Anthropic's pricing page as of November 20, 2026, with the full price table and a worked example.
The Claude API charges per million tokens (MTok), with separate prices for input and output. As of November 20, 2026, Anthropic lists Claude Haiku 5.5 at $0.10 input and $0.50 output per MTok, Claude Sonnet 5.5 at $2 and $10, Claude Opus 5.5 at $4 and $20, and Claude Fable 5.2 at $8 and $40. The Batch API takes 50% off both, and prompt cache hits cost a small fraction of the input price. Put your own numbers into the calculator below to see the cost per request, per day and per month.
- Per request
- -
- Per day
- -
- Per month
- -
Prices as of November 20, 2026, from Anthropic's pricing page. US dollars, Claude API (first party), standard global routing; taxes, web search and other tool fees not included.
The calculator runs in your browser and sends nothing anywhere. It uses the same prices as the table below, and the buttons under the result show the same usage on the four latest models.
Claude API prices per million tokens
Every model on Anthropic's pricing page, in US dollars per million tokens. Cache writes are what you pay the first time a prompt prefix is stored, for a 5-minute or a 1-hour cache. Cache hits are what you pay when a later request reads it; Anthropic bills cache refreshes at the same price. Batch prices are half the standard input and output prices.
| Model | Input | Output | Cache write (5 min) | Cache write (1 hour) | Cache hit | Batch input | Batch output |
|---|---|---|---|---|---|---|---|
| Fable 5.2 | $8 | $40 | $10 | $16 | $0.20 | $4 | $20 |
| Opus 5.5 | $4 | $20 | $5 | $8 | $0.20 | $2 | $10 |
| Sonnet 5.5 | $2 | $10 | $2.50 | $4 | $0.10 | $1 | $5 |
| Haiku 5.5 (prompt up to 100K tokens) | $0.10 | $0.50 | $0.125 | $0.20 | $0.01 | $0.05 | $0.25 |
| Haiku 5.5 (prompt over 100K tokens) | $0.50 | $2.50 | $0.625 | $1 | $0.05 | $0.25 | $1.25 |
| Mythos 5.2 | $8 | $40 | $10 | $16 | $0.20 | $4 | $20 |
| Fable 5.1 | $10 | $50 | $12.50 | $20 | $0.25 | $5 | $25 |
| Mythos 5.1 | $10 | $50 | $12.50 | $20 | $0.25 | $5 | $25 |
| Fable 5 | $10 | $50 | $12.50 | $20 | $1 | $5 | $25 |
| Mythos 5 | $10 | $50 | $12.50 | $20 | $1 | $5 | $25 |
| Opus 5 | $5 | $25 | $6.25 | $10 | $0.50 | $2.50 | $12.50 |
| Opus 4.8 | $5 | $25 | $6.25 | $10 | $0.50 | $2.50 | $12.50 |
| Opus 4.7 | $5 | $25 | $6.25 | $10 | $0.50 | $2.50 | $12.50 |
| Opus 4.6 | $5 | $25 | $6.25 | $10 | $0.50 | $2.50 | $12.50 |
| Opus 4.5 | $5 | $25 | $6.25 | $10 | $0.50 | $2.50 | $12.50 |
| Opus 4.1 | $15 | $75 | $18.75 | $30 | $1.50 | $7.50 | $37.50 |
| Opus 4 | $15 | $75 | $18.75 | $30 | $1.50 | $7.50 | $37.50 |
| Sonnet 5 | $2 | $10 | $2.50 | $4 | $0.20 | $1 | $5 |
| Sonnet 4.6 | $3 | $15 | $3.75 | $6 | $0.30 | $1.50 | $7.50 |
| Sonnet 4.5 | $3 | $15 | $3.75 | $6 | $0.30 | $1.50 | $7.50 |
| Sonnet 4 | $3 | $15 | $3.75 | $6 | $0.30 | $1.50 | $7.50 |
| Haiku 4.5 | $1 | $5 | $1.25 | $2 | $0.10 | $0.50 | $2.50 |
How the math works
One request costs its input tokens times the input price plus its output tokens times the output price, each divided by one million. Input means the whole prompt: system prompt, conversation history, documents and tool definitions all count, which is why input usually dominates the token count. The calculator then multiplies by requests per day, and by 30 for a month.
With prompt caching, the cached share of the input is billed at the cache-hit price on most requests. On the requests that store the cache, that share is billed at the write price instead: 1.25 times the input price for a 5-minute cache, 2 times for a 1-hour cache. The Batch API halves every price, and Anthropic says the caching multipliers stack with the batch discount. Claude Haiku 5.5 is priced by prompt length: a request whose prompt is over 100,000 tokens, counting cache reads and writes, pays the higher price set for the whole request.
Worked example: a support assistant on Claude Sonnet 5.5 sends 5,000 requests a day. Each request has 10,000 input tokens, 8,000 of them a fixed system prompt and help-center text kept in a 5-minute cache, and returns 1,000 output tokens. The cache is written 24 times a day.
- Uncached input: 2,000 tokens x 5,000 requests = 10 million tokens x $2 = $20.00 a day
- Cache hits: 8,000 tokens x 4,976 requests = 39.8 million tokens x $0.10 = $3.98 a day
- Cache writes: 8,000 tokens x 24 writes = 192,000 tokens x $2.50 = $0.48 a day
- Output: 1,000 tokens x 5,000 requests = 5 million tokens x $10 = $50.00 a day
- Total: $74.46 a day, about $0.0149 per request, $2,233.82 for a 30-day month
Without caching, the same workload costs $150 a day, or $4,500 a month, so the cache roughly halves the bill. Output is now the largest line, and caching cannot touch it. The same usage costs about $4,468 a month on Opus 5.5, $8,696 on Fable 5.2 and $118 on Haiku 5.5.
How to cut your Claude API bill
- Cache the parts that repeat. Anthropic says a 5-minute cache pays for itself after one cache read and a 1-hour cache after two. Hits are cheapest on the newest models: 0.025 times the input price on Fable 5.2, 0.05 times on Opus 5.5 and Sonnet 5.5, and 0.1 times on most other models.
- Send work that can wait through the Batch API. It is 50% off input and output tokens, and requests are processed asynchronously, so it suits evaluations, backfills and bulk classification rather than chat.
- Match the model to the job. Anthropic describes Haiku 5.5 as built for high-volume, latency-sensitive tasks such as classification, extraction and routing; it costs a twentieth of Sonnet 5.5 per token for prompts up to 100,000 tokens.
- Keep Haiku 5.5 prompts at or under 100,000 tokens where you can. Above that, input and output both cost five times as much for that request.
- Compare token counts across model generations, not only prices. Anthropic says Claude 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text than Sonnet 4.6 and earlier.
What the calculator leaves out
It covers standard token prices only. Fast mode costs more (Opus 5.5 is $8 input and $40 output per MTok in fast mode) and does not work with the Batch API. US-only inference on Claude 4.6 and later models multiplies every price by 1.1. Server-side tools such as web search are billed per use on top of tokens, and Amazon Bedrock and Google-managed Google Cloud publish their own price lists.
Using Claude Code instead of the API
If you run Claude Code on a Claude subscription rather than an API key, plan limits matter more than token prices. These two guides cover that side:
| Guide | What it answers |
|---|---|
| Claude Code pricing: plans and API costs | Which Claude plans include Claude Code, and what it costs at API rates |
| Claude Code usage limits explained | How the 5-hour and weekly caps work on Pro, Max and Ultra |
Related on claudefox
- Fable 5.2 pricing drops to $8/$40 per million tokens
- Claude Code pricing (November 2026): plans and API costs
- Reminder: Claude Sonnet 4.5 retires, and Opus 4.8 fast mode is removed, on November 30
- Gemini 4 Carbon vs Claude Opus 5.5 and Fable 5.2: benchmarks and pricing compared
Source: Pricing (Claude Platform docs), Claude plans and pricing (claude.com)