LLM APIs & Inference
Large language models you call from your code. Priced per token, with input and output billed separately.
Monthly cost as you grow
For a commercial project, using the cheapest plan that fits at each size and our default usage averages. How we estimate
| Provider | 100 users | 1,000 users | 10,000 users | 100,000 users |
|---|---|---|---|---|
| Anthropic Claude API | $0.40Pay as you go | $4.00Pay as you go | $40.00Pay as you go | $400Pay as you go |
| Google Gemini API | $1.70Pay as you go | $17.00Pay as you go | $170Pay as you go | $1,700Pay as you go |
| Groq API | $0Free | $0Free | $26.25Developer (pay as you go) | $263Developer (pay as you go) |
| Mistral AI API | $0.20Pay as you go | $2.00Pay as you go | $20.00Pay as you go | $200Pay as you go |
| OpenAI API | $0.40Pay as you go | $4.00Pay as you go | $40.00Pay as you go | $400Pay as you go |
Free tiers and their catches
Anthropic Claude API
No free plan.
Google Gemini API
No free plan.
Groq API
Free: Groq's rate limits page (https://console.groq.com/docs/rate-limits) shows Free plan limits by default: GPT OSS 20B, GPT OSS 120B and Qwen3.8 27B each allow 30 requests a minute and 1,000 a day. Over a limit the API returns a 429 error.
- Free stops at its limits rather than billing more.
- Free allows 1,000 requests per day (ai requests).
Mistral AI API
No free plan.
OpenAI API
No free plan.