Marketing copy, product descriptions, emails and SEO drafts: short instruction input, long creative output.
Input / request
800
Output / request
1,200
Cacheable input
30%
Requests / user / day
20
Cost by model
| Model | Per request | Per user / month | 500 users / month | vs cheapest |
|---|---|---|---|---|
| bestMinistral 3 (8B) | $0.000268 | $0.1606 | $80.28 | — |
| Gemini 2.5 Flash-Lite | $0.000538 | $0.323 | $161.52 | 2.0× |
| Mistral Small 4 | $0.000808 | $0.4846 | $242.28 | 3.0× |
| Codestral | $0.001255 | $0.7531 | $376.56 | 4.7× |
| GPT-5.6 Luna | $0.001557 | $0.9341 | $467.04 | 5.8× |
| GPT-5.4 nano | $0.001617 | $0.9701 | $485.04 | 6.0× |
| DeepSeek V4 Flash | $0.001834 | $1.10 | $550.128 | 6.9× |
| Mistral Large 3 | $0.002092 | $1.255 | $627.60 | 7.8× |
| Gemini 3.5 Flash-Lite | $0.003175 | $1.905 | $952.56 | 11.9× |
| Gemini 2.5 Flash | $0.003175 | $1.905 | $952.56 | 11.9× |
| Grok 4.3 | $0.003748 | $2.249 | $1.1k | 14.0× |
| DeepSeek V4 Pro | $0.005502 | $3.301 | $1.7k | 20.6× |
| GPT-5.4 mini | $0.005838 | $3.503 | $1.8k | 21.8× |
| o4-mini | $0.005962 | $3.577 | $1.8k | 22.3× |
| GLM 5.2 | $0.006098 | $3.659 | $1.8k | 22.8× |
| Muse Spark | $0.0061 | $3.66 | $1.8k | 22.8× |
| Claude Haiku 4.5 | $0.006584 | $3.95 | $2k | 24.6× |
| Grok 4.5 | $0.008392 | $5.035 | $2.5k | 31.4× |
| Grok 4.6 | $0.00844 | $5.064 | $2.5k | 31.5× |
| Gemini 3.6 Flash | $0.009876 | $5.926 | $3k | 36.9× |
| Mistral Medium 3.5 | $0.009876 | $5.926 | $3k | 36.9× |
| o3 | $0.0108 | $6.504 | $3.3k | 40.5× |
| GPT-4.1 | $0.0108 | $6.504 | $3.3k | 40.5× |
| Gemini 3.5 Flash | $0.0117 | $7.006 | $3.5k | 43.6× |
| GPT-5.1 | $0.0127 | $7.638 | $3.8k | 47.6× |
| Gemini 2.5 Pro | $0.0127 | $7.638 | $3.8k | 47.6× |
| Claude Sonnet 5 | $0.0132 | $7.901 | $4k | 49.2× |
| GPT-5.6 Terra | $0.0156 | $9.341 | $4.7k | 58.2× |
| Gemini 3.1 Pro | $0.0156 | $9.341 | $4.7k | 58.2× |
| GPT-5.2 | $0.0178 | $10.693 | $5.3k | 66.6× |
| GPT-5.4 | $0.0195 | $11.676 | $5.8k | 72.7× |
| Claude Sonnet 4.5 | $0.0198 | $11.851 | $5.9k | 73.8× |
| Kimi K3 | $0.0198 | $11.851 | $5.9k | 73.8× |
| Claude Opus 5 | $0.0329 | $19.752 | $9.9k | 123.0× |
| Claude Opus 4.5 | $0.0329 | $19.752 | $9.9k | 123.0× |
| GPT-5.6 Sol | $0.0389 | $23.352 | $11.7k | 145.4× |
| GPT-5.5 | $0.0389 | $23.352 | $11.7k | 145.4× |
| Claude Fable 5 | $0.0658 | $39.504 | $19.8k | 246.0× |
Assumes the typical cacheable share (30% of input at cached rates where published). 20 requests/user/day. Adjust everything in the monthly calculator.
How to spend less
Related workflow costs
FAQ
Using a typical profile of 800 input and 1,200 output tokens with 30% cacheable input: from $0.000268 on Ministral 3 (8B) up to $0.0658 on Claude Fable 5. See the table for every model.
At 20 requests per user per day and the typical token profile, budget from $0.1606 per active user per month on Ministral 3 (8B). A team of 500 active users lands around $80.28/month at that tier. Model your exact numbers in the monthly cost calculator.
Output-heavy workloads favor models with a low output price, not a low input price. Batch similar generation tasks with shared style prompts to exploit caching. Draft with a cheap tier, refine the winners with a premium model.