Free tool
LLM API Cost Calculator
Estimate LLM API cost per call and per month from verified prices, with uncached input, cache reads, cache writes and output priced separately.
Anthropic
Meta
OpenAI
Only models with verified prices are listed. A model is added after we check the vendor’s price page.
Everything you send: system prompt, history, tools and user text.
Share of input tokens read from cache — not share of requests.
Includes reasoning tokens — don't add them twice.
= 2,000 uncached + 0 cache-read tokens per call. Cache writes: assumed 0 — switch to Detailed to add them.
Volume
API calls — not users or conversations.
The period your total covers.
GPT-6.1 Sol
OpenAI API · Standard · gpt-6.1-sol
1,000 API calls/day × 30 days
- Uncached input16.7%
- Cache read0.0%
- Cache write0.0%
- Output83.3%
- Standard rate (price threshold 272,000 prompt tokens) — 2,000 prompt tokens per call.
- Excludes tax. Totals are summed before rounding.
Compare
Same usage on other models
Uses the tokens per call and volume above. Prices are list prices for each vendor’s own API, standard tier.
| Model | List price / 1M · in · out | Per call | Period · 30 days | vs GPT-6.1 Sol |
|---|---|---|---|---|
| GPT-6.1 SolSelected · OpenAI | $2.00 · $10.00 | $0.02400 | $720.00 | Selected |
| Claude Opus 5.5Anthropic | $4.00 · $20.00 | $0.04800 | $1,440.00 | +$720.00 · +100.0% |
| Claude Sonnet 5.5Anthropic | $2.00 · $10.00 | $0.02400 | $720.00 | ±$0.00 · 0.0% |
| Claude Haiku 5.5Anthropic | $0.10 · $0.50 | $0.001200 | $36.00 | −$684.00 · −95.0% |
| Muse WonderPreviewMeta | $4.00 · $20.00 | $0.04800 | $1,440.00 | +$720.00 · +100.0% |
| Muse Spark 1.3Meta | $1.25 · $4.25 | $0.01100 | $330.00 | −$390.00 · −54.2% |
| GPT-6 AstraOpenAI | $10.00 · $50.00 | $0.1200 | $3,600.00 | +$2,880.00 · +400.0% |
| GPT-6 LunaOpenAI | $0.10 · $0.50 | $0.001200 | $36.00 | −$684.00 · −95.0% |
- Default: cache reads are priced as new input on other models — a cache hit rate doesn’t carry over to a model you haven’t run.
- Each model’s tokenizer is different, so the same text has a different token count. Equal token counts here show a price scenario only.
- Tool calls such as web search aren’t included in any row.
Method
How the calculator prices a call
Each call is split into parts that never overlap: uncached input, cache reads, short and long cache writes, and output (reasoning tokens included). Each part is priced at its own rate, per million tokens.
When a model charges more above a prompt-length threshold, every request is tiered by its own length. We never average long and short requests into one size.
What isn’t included
- Tool calls such as web search, images, audio and other modalities.
- Taxes, credits, discounts and negotiated prices.
- Cloud resellers and aggregators — prices are for each vendor’s own API.
Limits and rounding
A call whose input plus output is over the model’s context window gets no estimate. Per-call amounts show four significant digits; period totals show cents. Totals are added up at full precision and rounded once.