AI API cost calculator
Estimate and compare monthly API costs for GPT, Claude, Gemini, DeepSeek, Grok and more from input/output tokens and request volume. Find the cheapest model for your use case.
A. What this tool does
Enter your usage (token counts and request volume) and it automatically computes and compares the monthly cost of 13 major models. Shown in USD and an approximate JPY figure (US$1 ≈ ¥155).
Quick presets:
Monthly = (input ÷ 1M × input rate + output ÷ 1M × output rate) × requests × 30 days. Token counts are approximate (about 1 character = 1 token for Japanese).
| Model | Input $/1M | Output $/1M | Monthly (USD) | Monthly (approx. ¥) |
|---|---|---|---|---|
| OpGPT-6 Luna OpenAI · Low cost, lightweight | $0.1 | $0.5 | $2.10 | ¥325 |
| AnClaude Haiku 5.5 Anthropic · up to 100K tokens | $0.1 | $0.5 | $2.10 | ¥325 |
| DeDeepSeek V4.1 Flash DeepSeek · Peak-hour price (half price off-peak) | $0.3 | $1.2 | $5.40 | ¥837 |
| GoGemini 3.5 Flash-Lite Google · Low cost, lightweight | $0.3 | $2.5 | $9.30 | ¥1,442 |
| xAGrok 4.3 xAI · Low cost | $1.25 | $2.5 | $15.00 | ¥2,325 |
| GoGemini 3.8 Flash Google · Rises to $1.5/$7.5 from 2027-01-01 | $0.75 | $3.75 | $15.75 | ¥2,441 |
| DeDeepSeek V4 Pro DeepSeek · Peak-hour price (half price off-peak) | $1.32 | $3.96 | $19.80 | ¥3,069 |
| xAGrok 4.7 xAI · up to 200K tokens | $2 | $6 | $30.00 | ¥4,650 |
| OpGPT-6.1 Sol OpenAI · Balanced | $2 | $10 | $42.00 | ¥6,510 |
| AnClaude Sonnet 5.5 Anthropic · Balanced | $2 | $10 | $42.00 | ¥6,510 |
| GoGemini 4 Argon Google · Rises to $4/$20 from 2026-12-01 | $2 | $10 | $42.00 | ¥6,510 |
| AnClaude Opus 5.5 Anthropic · High performance | $4 | $20 | $84.00 | ¥13,020 |
| OpGPT-6 Astra OpenAI · up to 272K tokens | $10 | $50 | $210 | ¥32,550 |
Source: Official API pricing from each provider ／ Captured: 2026-10-17 ／ Reference values; check each official site for the latest (prices change)
Pricing sources (official)
Frequently asked questions
How is AI API pricing determined?
Most LLMs set a per-1M-token rate for "input tokens" and "output tokens" separately and bill on a pay-as-you-go basis. The output rate is usually higher than the input rate.
Which AI API is the cheapest?
On a per-rate basis, OpenAI's GPT-6 Luna and similar models (input $0.1/1M) are in the low-price tier. Still, you should choose by performance and use case — use the simulator above to compare monthly costs for your actual usage.
What's a rough guide for token counts?
As a rough guide, English is about 1.3 tokens per word and Japanese is about 1 token per character. The longer the document, the more tokens — and the higher the cost.
Related pages
⚠ Reference values based on public data and editorial research; not a complete count of actual users.