1. 利用シナリオ
入力トークン送信するテキスト量
出力トークンAIが生成する応答量
推論トークン
AIの思考プロセス（一部モデルのみ）API 回数リクエストの総回数
プロンプトキャッシュ繰り返しリクエスト時に入力の80%を再利用してコスト削減
速度を表示
プロンプトを直接入力 (任意)
2. コストシミュレーション
| モデル | 合計最小 | 合計最大 | 合計最小~合計最大 | 単価 |
|---|---|---|---|---|
 Llama 4 Scout | $1.32 | $2.76 | $1.32~2.76 | $0.080/1M |
 GPT OSS 120B | $1.37 | $2.28 | $1.37~2.28 | $0.039/1M |
Llama 4 Maverick | $2.61 | $5.49 | $2.61~5.49 | $0.150/1M |
GPT-5 Nano | $2.79 | $4.71 | $2.79~4.71 | $0.050/1M |
 Gemma 4 31B | $2.90 | $4.72 | $2.90~4.72 | $0.130/1M |
 Nemotron 3 Super | $3.24 | $5.40 | $3.24~5.40 | $0.090/1M |
 DeepSeek V3.2 | $3.25 | $5.07 | $3.25~5.07 | $0.252/1M |
 Longcat Flash Chat | $3.48 | $7.32 | $3.48~7.32 | $0.200/1M |
 Grok 4.1 Fast | $3.90 | $6.30 | $3.90~6.30 | $0.200/1M |
Grok 4.1 Fast (Reasoning) | $3.90 | $6.30 | $3.90~6.30 | $0.200/1M |
 Mistral Small 4 | $4.41 | $7.29 | $4.41~7.29 | $0.150/1M |
 ERNIE 4.5 300B A47B | $4.80 | $10.08 | $4.80~10.08 | $0.280/1M |
 MiniMax M2.7 | $5.22 | $10.98 | $5.22~10.98 | $0.300/1M |
 K-EXAONE | $5.88 | $9.72 | $5.88~9.72 | $0.200/1M |
 Trinity Large Thinking | $6.27 | $10.35 | $6.27~10.35 | $0.220/1M |
MiniMax M2.5 | $8.37 | $14.13 | $8.37~14.13 | $0.150/1M |
GPT-5.4 Nano | $8.85 | $14.85 | $8.85~14.85 | $0.200/1M |
Gemini 3.1 Flash Lite | $10.65 | $17.85 | $10.65~17.85 | $0.250/1M |
 Qwen3.6 Plus | $13.84 | $23.20 | $13.84~23.20 | $0.325/1M |
GPT-5 Mini | $13.95 | $23.55 | $13.95~23.55 | $0.250/1M |
 Kimi K2.5 | $14.52 | $24.12 | $14.52~24.12 | $0.440/1M |
 GLM-5 | $15.68 | $25.66 | $15.68~25.66 | $0.650/1M |
Qwen3.5 397B A17B | $16.61 | $27.85 | $16.61~27.85 | $0.390/1M |
Gemini 2.5 Flash | $17.40 | $29.40 | $17.40~29.40 | $0.300/1M |
Gemini 2.5 Flash Lite | $17.40 | $29.40 | $17.40~29.40 | $0.300/1M |
 Nova 2 Lite | $17.40 | $29.40 | $17.40~29.40 | $0.300/1M |
Gemini 3 Flash | $21.30 | $35.70 | $21.30~35.70 | $0.500/1M |
 MiMo-V2-Pro | $22.80 | $37.20 | $22.80~37.20 | $1.00/1M |
GLM-5.1 | $26.25 | $43.05 | $26.25~43.05 | $1.05/1M |
Grok 4.20 | $27.60 | $56.40 | $27.60~56.40 | $2.00/1M |
GPT-5.4 Mini | $31.95 | $53.55 | $31.95~53.55 | $0.750/1M |
GPT-4.1 | $34.80 | $73.20 | $34.80~73.20 | $2.00/1M |
 Claude Haiku 4.5 | $36.00 | $60.00 | $36.00~60.00 | $1.00/1M |
Grok 4.20 (Reasoning) | $45.60 | $74.40 | $45.60~74.40 | $2.00/1M |
Gemini 2.5 Pro | $69.75 | $117.75 | $69.75~117.75 | $1.25/1M |
GPT-5 | $69.75 | $117.75 | $69.75~117.75 | $1.25/1M |
Gemini 3.1 Pro | $85.20 | $142.80 | $85.20~142.80 | $2.00/1M |
GPT-5.4 | $106.50 | $178.50 | $106.50~178.50 | $2.50/1M |
Claude Sonnet 4 | $108.00 | $180.00 | $108.00~180.00 | $3.00/1M |
Claude Sonnet 4.5 | $108.00 | $180.00 | $108.00~180.00 | $3.00/1M |
Claude Sonnet 4.6 | $108.00 | $180.00 | $108.00~180.00 | $3.00/1M |
Claude Opus 4.5 | $180.00 | $300.00 | $180.00~300.00 | $5.00/1M |
Claude Opus 4.6 | $180.00 | $300.00 | $180.00~300.00 | $5.00/1M |
Claude Opus 4.7 | $180.00 | $300.00 | $180.00~300.00 | $5.00/1M |
Claude Opus 4 | $540.00 | $900.00 | $540.00~900.00 | $15.00/1M |
Claude Opus 4.1 | $540.00 | $900.00 | $540.00~900.00 | $15.00/1M |
GPT-5.4 Pro | $1278.00 | $2142.00 | $1278.00~2142.00 | $30.00/1M |
4. シミュレーション要約
最低コストモデル
Llama 4 Scout
$1.32 /3,000回
最高性能モデル
Claude Opus 4.7
$180.00 /3,000回
算出基準
入力トークン: 1,000
出力トークン: 1,200 ~ 2,800 (±40%)
推論トークン: 1,000
使用量: 3,000 回
トークンプリセットはシナリオ別の統計的平均値です。実際のトークン数はプロンプト内容により異なります。ReasoningトークンはExtended Thinkingをサポートするモデルにのみ適用されます。
単価最終更新: 2026년 4월 22일