AI API Pricing Comparison Table
Compare live per-token pricing for 111 models from 8 providers, including OpenAI, Claude, Gemini, Grok, DeepSeek, and Mistral. If you searched for AI API pricing, API cost, API price comparison, or an AI API pricing calculator, this page is the shortest route from raw rates to a usable budget estimate.
Popular AI API pricing paths
Use the full pricing table for raw model rates, the cost calculator for budget math, and provider pages when you already know the stack you want to price.
Project cost per request, cost per 1,000 calls, and monthly spend from your own token volumes.
Compare two models side by side when cheap is not enough and quality still matters.
Jump straight to OpenAI model prices, token rates, and calculator paths.
See Anthropic pricing and the cheapest Claude options without digging through the full table.
36 models, token pricing, and cheapest picks by tier.
17 models, token pricing, and cheapest picks by tier.
18 models, token pricing, and cheapest picks by tier.
10 models, token pricing, and cheapest picks by tier.
6 models, token pricing, and cheapest picks by tier.
14 models, token pricing, and cheapest picks by tier.
How AI API Pricing Works
💡 Input vs Output Tokens
AI APIs charge separately for input tokens (your prompt, context, instructions) and output tokens (the model's response). Output tokens are typically 2-8× more expensive because they require more computation.
📊 Cost Per 1K Requests
The “~Cost/1K req” column estimates costs for 1,000 typical API calls, assuming an average of 750 input tokens and 250 output tokens per request. Your actual costs will vary based on your prompt and response lengths.
🏷️ Model Categories
Flagship models offer the best quality. Reasoning models excel at complex tasks. Budget models balance cost and quality. Choose based on your quality requirements and budget.
💰 Saving Tips
Use prompt caching to reduce repeated input costs. Try batch APIs for non-real-time work (up to 50% off). Start with cheaper models and only upgrade when quality demands it.
More Tools
Frequently Asked Questions
What is the cheapest AI API in 2026?▼
GPT-5 nano offers the lowest input pricing at $0.05/M tokens. However, the cheapest option depends on your use case — consider output costs, context window, and capabilities too. Use the table above to sort by the metric that matters most to you.
How is AI API pricing calculated?▼
AI APIs charge per token (roughly ¾ of a word). Pricing is listed per million tokens, with separate rates for input (your prompt) and output (the response). Total cost = (input tokens × input rate) + (output tokens × output rate).
What does “per million tokens” mean?▼
Tokens are the basic units AI models process. One million tokens is roughly 750,000 words. Pricing is quoted per million tokens to make small per-token costs easier to compare.
Which AI provider is best for cost-sensitive applications?▼
For cost-sensitive apps, consider budget-tier models like DeepSeek, Mistral's smaller models, or Google's Gemini Flash. These offer strong performance at a fraction of flagship pricing. Use our batch calculator to estimate costs for your specific workload.