Compare token pricing across OpenAI, Anthropic, Google, Mistral, and Meta AI APIs. Enter your expected monthly usage and see cost breakdowns per model, sorted cheapest-first.
Prefer the main AI Wins product? Visit aiwins.news
How it works
Input your expected monthly token counts for both input and output, or pick a preset to get started quickly.
See a side-by-side cost breakdown for 18+ models across five providers, sorted from cheapest to most expensive.
Filter by provider, sort by total cost, and identify the model that delivers the best performance for your budget.
Calculator
Tokens you send to the API (prompts, context, instructions)
Tokens the model generates in response (completions)
Quick presets
Filter by provider
Results
Showing 18 of 18 models
| Model | Provider | Input Cost | Output Cost | Total / Month | Context |
|---|---|---|---|---|---|
Ministral 8BCheapest | Mistral | $0.10 | $0.05 | $0.15 | 128K |
Gemini 1.5 Flash | $0.07 | $0.15 | $0.22 | 1M | |
Llama 3.1 8B | Meta | $0.18 | $0.09 | $0.27 | 128K |
Gemini 2.0 Flash | $0.10 | $0.20 | $0.30 | 1M | |
GPT-4o Mini | OpenAI | $0.15 | $0.30 | $0.45 | 128K |
Mistral Small | Mistral | $0.20 | $0.30 | $0.50 | 128K |
Llama 3.1 70B | Meta | $0.88 | $0.44 | $1.32 | 128K |
Claude 3.5 Haiku | Anthropic | $0.80 | $2.00 | $2.80 | 200K |
o3-mini | OpenAI | $1.10 | $2.20 | $3.30 | 200K |
Gemini 1.5 ProPopular | $1.25 | $2.50 | $3.75 | 2M | |
Mistral LargePopular | Mistral | $2.00 | $3.00 | $5.00 | 128K |
Llama 3.1 405B | Meta | $3.50 | $1.75 | $5.25 | 128K |
GPT-4oPopular | OpenAI | $2.50 | $5.00 | $7.50 | 128K |
o1-mini | OpenAI | $3.00 | $6.00 | $9.00 | 128K |
Claude 3.5 SonnetPopular | Anthropic | $3.00 | $7.50 | $10.50 | 200K |
GPT-4 Turbo | OpenAI | $10.00 | $15.00 | $25.00 | 128K |
o1 | OpenAI | $15.00 | $30.00 | $45.00 | 200K |
Claude 3 Opus | Anthropic | $15.00 | $37.50 | $52.50 | 200K |
Visual comparison
FAQ
AI API providers charge based on the number of tokens processed. Tokens are pieces of text (roughly 4 characters or 0.75 words each). You pay separately for input tokens (what you send to the model) and output tokens (what the model generates). Multiply your token count by the per-token rate to get the total cost.
Input tokens are the text you send to the API, including your prompt, system instructions, and any context. Output tokens are the text the model generates in response. Output tokens typically cost 2-5x more than input tokens because generation requires more compute.
For simple tasks, Google Gemini 1.5 Flash and Mistral's Ministral 8B offer some of the lowest per-token rates. For complex reasoning, GPT-4o Mini and Claude 3.5 Haiku provide strong performance at budget-friendly prices. The cheapest option depends on your quality requirements and use case.
A typical API call uses 500-2,000 input tokens and generates 200-1,000 output tokens. Multiply by your expected number of requests per month. For example, 10,000 requests at 1,000 input and 500 output tokens each equals 10M input tokens and 5M output tokens per month.
Yes, most providers offer free credits or trial tiers. Google provides a generous free tier for Gemini, OpenAI gives new accounts starter credits, and Anthropic offers trial credits. Meta's Llama models are open-source and free to self-host, though hosted inference providers charge for compute.
Beyond token pricing, costs can vary based on rate limits (needing a higher tier for more throughput), fine-tuning fees, image or audio inputs, batch vs real-time pricing, and whether you use prompt caching. Some providers also offer volume discounts for committed usage.
Related tools
Browse 20+ popular AI tools across six categories in a pre-populated, filterable chart. Compare pricing, features, API access, and standout capabilities at a glance.
Open toolEvaluate whether your team is ready to adopt AI tools with a practical score across data, talent, and strategy dimensions.
Open toolAnalyze your AI prompts for common weaknesses and get an improved version instantly. Checks specificity, role assignment, examples, constraints, and more.
Open tool