Translating content across languages. Token counts vary by language — CJK scripts tokenize differently from Latin scripts, which changes cost per word.
Input / request
5,000
Output / request
5,500
Cacheable input
15%
Requests / user / day
15
Cost by model
| Model | Per request | Per user / month | 500 users / month | vs cheapest |
|---|---|---|---|---|
| bestMinistral 3 (8B) | $0.001474 | $0.6632 | $331.594 | — |
| Gemini 2.5 Flash-Lite | $0.002633 | $1.185 | $592.313 | 1.8× |
| Mistral Small 4 | $0.003949 | $1.777 | $888.469 | 2.7× |
| Codestral | $0.006248 | $2.811 | $1.4k | 4.2× |
| GPT-5.6 Luna | $0.007465 | $3.359 | $1.7k | 5.1× |
| GPT-5.4 nano | $0.00774 | $3.483 | $1.7k | 5.3× |
| DeepSeek V4 Flash | $0.009141 | $4.113 | $2.1k | 6.2× |
| Mistral Large 3 | $0.0104 | $4.686 | $2.3k | 7.1× |
| Gemini 3.5 Flash-Lite | $0.015 | $6.771 | $3.4k | 10.2× |
| Gemini 2.5 Flash | $0.015 | $6.771 | $3.4k | 10.2× |
| Grok 4.3 | $0.0192 | $8.646 | $4.3k | 13.0× |
| DeepSeek V4 Pro | $0.0274 | $12.34 | $6.2k | 18.6× |
| GPT-5.4 mini | $0.028 | $12.597 | $6.3k | 19.0× |
| o4-mini | $0.0291 | $13.087 | $6.5k | 19.7× |
| Muse Spark | $0.0296 | $13.331 | $6.7k | 20.1× |
| GLM 5.2 | $0.0303 | $13.615 | $6.8k | 20.5× |
| Claude Haiku 4.5 | $0.0318 | $14.321 | $7.2k | 21.6× |
| Grok 4.5 | $0.0417 | $18.776 | $9.4k | 28.3× |
| Grok 4.6 | $0.0419 | $18.844 | $9.4k | 28.4× |
| Gemini 3.6 Flash | $0.0477 | $21.482 | $10.7k | 32.4× |
| Mistral Medium 3.5 | $0.0477 | $21.482 | $10.7k | 32.4× |
| o3 | $0.0529 | $23.794 | $11.9k | 35.9× |
| GPT-4.1 | $0.0529 | $23.794 | $11.9k | 35.9× |
| Gemini 3.5 Flash | $0.056 | $25.194 | $12.6k | 38.0× |
| GPT-5.1 | $0.0604 | $27.183 | $13.6k | 41.0× |
| Gemini 2.5 Pro | $0.0604 | $27.183 | $13.6k | 41.0× |
| Claude Sonnet 5 | $0.0637 | $28.643 | $14.3k | 43.2× |
| GPT-5.6 Terra | $0.0747 | $33.593 | $16.8k | 50.7× |
| Gemini 3.1 Pro | $0.0747 | $33.593 | $16.8k | 50.7× |
| GPT-5.2 | $0.0846 | $38.056 | $19k | 57.4× |
| GPT-5.4 | $0.0933 | $41.991 | $21k | 63.3× |
| Claude Sonnet 4.5 | $0.0955 | $42.964 | $21.5k | 64.8× |
| Kimi K3 | $0.0955 | $42.964 | $21.5k | 64.8× |
| Claude Opus 5 | $0.1591 | $71.606 | $35.8k | 108.0× |
| Claude Opus 4.5 | $0.1591 | $71.606 | $35.8k | 108.0× |
| GPT-5.6 Sol | $0.1866 | $83.981 | $42k | 126.6× |
| GPT-5.5 | $0.1866 | $83.981 | $42k | 126.6× |
| Claude Fable 5 | $0.3182 | $143.212 | $71.6k | 215.9× |
Assumes the typical cacheable share (15% of input at cached rates where published). 15 requests/user/day. Adjust everything in the monthly calculator.
How to spend less
Related workflow costs
FAQ
Using a typical profile of 5,000 input and 5,500 output tokens with 15% cacheable input: from $0.001474 on Ministral 3 (8B) up to $0.3182 on Claude Fable 5. See the table for every model.
At 15 requests per user per day and the typical token profile, budget from $0.6632 per active user per month on Ministral 3 (8B). A team of 500 active users lands around $331.594/month at that tier. Model your exact numbers in the monthly cost calculator.
Output length ≈ input length; budget both sides of the request. Japanese and Chinese text typically costs more per word than English due to tokenization. Cache translation memories and glossaries embedded in the prompt.