How much does the Claude API cost per month?

It depends on your usage. Claude Opus 4.8 costs $5 per million input tokens and $25 per million output. Claude Haiku 4.5 costs $1/$5. Use this calculator: enter your monthly token count and it shows the exact monthly cost for every Claude tier and alternative.

What is the cheapest AI API in 2026?

DeepSeek V4-Flash is currently the cheapest commercial API at $0.14/$0.28 per million tokens. Llama 4 and Phi-4 are free if you self-host, but you pay for the compute.

How do I estimate my monthly AI token usage?

A typical chat message is about 500 input tokens and 1,200 output tokens. If you send 1,000 messages per day, that's ~15M input and 36M output tokens per month. Use the messages-per-day mode in this calculator for a quick estimate.

Are GPT-5 and Gemini pricing verified?

Yes. Pricing is taken from each provider's official pricing page and verified monthly. GPT-5 is $1.25/$10 per million tokens. GPT-5.5 is $5/$30. Gemini 3.5 Flash is $1.50/$9. See the data docs page for the full source list.

Cost Calculator · Prices verified June 2026

AI API cost calculator

Enter your monthly token usage or messages per day. benchr calculates the real cost for every model in the index and ranks them cheapest-first — no email, no sign-up.

Prices from official provider docs Calculations verified — never paid placements

Your usage

Monthly input tokens Tokens you send to the model per month

Monthly output tokens Tokens the model generates per month

Don't know your token count? Rough guide: one word ≈ 1.3 tokens. A long email ≈ 800 tokens. A 10-page document ≈ 6,000 tokens. Or try a preset above.

Enter usage above to see cost calculations.

Monthly cost

APIs cheapest-first · self-hosted listed separately

Loading…

Free/self-hosted models show $0 API cost — actual cost is GPU/cloud infrastructure. Pricing verified June 2026. See methodology for sources.

Also useful

→ Model rankings — capability + price together → Model recommender — guided questions → Price per use case — when cost changes everything → How to cut your token bill

Frequently asked questions

What's the cheapest AI API?

DeepSeek V4-Flash at $0.14/$0.28 per million tokens is the cheapest commercial API in this set. If you self-host Llama 4 or Phi-4, your API fee is $0 — you only pay for the compute. See the free coding models guide for the self-host trade-offs.

How much does Claude cost per month?

Depends on which tier. At 1M input + 500K output tokens per month: Haiku 4.5 ≈ $3.50, Sonnet 4.6 ≈ $10.50, Opus 4.7/4.8 ≈ $17.50. Enter your actual numbers above for the exact figure.

Does this include cached pricing?

Not yet — this calculator uses standard (non-cached) input pricing for simplicity. Most providers offer cache discounts of 50–90% on repeated context. If caching applies to your workload, your actual cost will be lower. Check the compare tool for per-model cached pricing.

Is the pricing up to date?

Prices are verified against official provider pricing pages and noted in the changelog. Provider prices can change without notice — verify before signing a contract.