Claude Token Cost Calculator
This Claude token cost calculator converts your input, output and cache-read tokens into a per-call estimate and a projected monthly API bill. Select a model and use the rate tier that matches your workload.
Cost calculator
Calculate one call, then project the same workload over a month.
- Input charge $4.00 /1M effective
- $0.40
- Uncached input
- $0.40
- Cache reads
- $0.00
- Cache writes
- Excluded
- Output charge $20.00 /1M
- $0.20
- Monthly calls
- 10
Claude Token Cost Calculator rates
USD per 1M tokens · standard ratesThese are the same five models and rates shown on the full Claude pricing page. Cache write below is the five-minute tier.
| Model | Input | Output | Cache read | Cache write |
|---|---|---|---|---|
| Claude Opus 5.5 | $4.00 | $20.00 | $0.20 | $5.00 |
| Claude Opus 5 | $5.00 | $25.00 | $0.50 | $6.25 |
| Claude Sonnet 5.5 | $2.00 | $10.00 | $0.20 | $2.50 |
| Claude Haiku 4.5 | $1.00 | $5.00 | $0.10 | $1.25 |
| Claude Fable 5.1 | $10.00 | $50.00 | $0.25 | $12.50 |
Pricing last verified . Batch, Fast and US-only use the same verified or derived tiers as the homepage; unavailable tiers are disabled rather than estimated.
How token cost is calculated
For the basic estimate, let I be input tokens, O output tokens and h the cache hit fraction. With rates quoted per million tokens:
Per call = (I × (1 − h) × input rate + I × h × cache-read rate + O × output rate) ÷ 1,000,000Monthly cost = per-call cost × calls per monthFor a Claude Opus 5.5 cost calculator example, use 1M input and 100K output tokens. This Claude token calculator applies exactly the same arithmetic as the homepage examples:
- Opus 5.5 · no cache
1 × $4.00 + 0.1 × $20.00 = $6.00- Opus 5.5 · 90% cache hit
0.1 × $4.00 + 0.9 × $0.20 + 0.1 × $20.00 = $2.58
Both examples exclude cache writes. Advanced cache budget uses the shared estimator: write share = h ÷ cached reads per write, replacing the same share of uncached input, and uses the selected five-minute or one-hour write rate. Writes plus reads must fit within the input workload.
What changes your bill
- Cache hit rate shifts input tokens from the standard rate to the cache-read rate: Opus 5.5 reads are $0.20 versus $4.00 per 1M, or 0.05×, before writes.
- Batch applies a 0.5× multiplier to standard input and output rates, a 50% discount for non-urgent jobs.
- Fast mode applies a 2× multiplier where the selected model has verified Fast rates.
- US-only inference applies a 1.1× multiplier where that tier is verified for the selected model.
- Output volume scales its charge directly: 2× output tokens means 2× output cost at the same rate, without cache-read savings.
Use this Anthropic token cost calculator to compare a fixed workload; the Claude rate tiers and worked examples explain the pricing assumptions behind those estimates.
Claude token calculator FAQ
How much does 1M Claude tokens cost?
Input and output tokens have separate rates. For Claude Opus 5.5, 1M uncached input tokens cost $4.00 and 1M output tokens cost $20.00 at standard rates; mixed workloads depend on the split.
Is there a free way to estimate Claude costs?
Yes. This Claude API cost calculator is free and runs locally in your browser without an API key or a model request. Enter your token counts and monthly calls to estimate both totals.
Does Claude charge for cached tokens?
Yes. Opus 5.5 cache reads cost $0.20 per 1M tokens, and five-minute cache writes cost $5.00 per 1M. The basic estimate excludes writes; open Advanced cache budget to include the existing write and reuse assumptions.
How do I calculate Claude Code token costs?
Use your API usage totals for input, output and cached input, select the model, and enter the same workload here. This is not a Claude Code token counter: it does not inspect your files or calculate a subscription bill from its monthly fee.
What is the cheapest Claude model per token?
Among the five models in this table, Claude Haiku 4.5 has the lowest standard input and output rates: $1.00 and $5.00 per 1M tokens. Compare quality on your task before choosing on price alone.
Can I combine cache savings with batch or fast mode?
This estimator does not combine caching with a non-standard tier, matching the homepage calculator. Switching to Batch, Fast or US-only sets cache hits to zero and excludes cache writes; tiers without verified rates are unavailable for that model.