OpenAI API Pricing (February 2026)
Current pricing for all OpenAI models. Prices are per 1 million tokens. Use our calculator to get exact costs for your prompts.
| Model | Input Price | Output Price | Context Window |
|---|---|---|---|
| GPT-5Flagship | $15.00 / 1M | $45.00 / 1M | 256K tokens |
| GPT-5.2 | $18.00 / 1M | $54.00 / 1M | 256K tokens |
| GPT-4oRecommended | $2.50 / 1M | $10.00 / 1M | 128K tokens |
| GPT-4o-miniBest Value | $0.15 / 1M | $0.60 / 1M | 128K tokens |
| o3 | $10.00 / 1M | $40.00 / 1M | 200K tokens |
| o3-mini | $1.10 / 1M | $4.40 / 1M | 200K tokens |
Pro Tip: Prompt Caching
OpenAI offers 50% discount on cached input tokens. If your system prompt stays the same across requests, you automatically get cheaper pricing on subsequent calls.
How to Count OpenAI Tokens
OpenAI uses the tiktoken library for tokenization. Here are the encoding types for each model:
Encoding by Model
- GPT-5, GPT-4o, GPT-4o-mini: cl100k_base encoding
- o3, o3-mini: cl100k_base encoding
- GPT-3.5-turbo: cl100k_base encoding
Python Code Example
import tiktoken
# For GPT-5, GPT-4o, GPT-4o-mini
enc = tiktoken.encoding_for_model("gpt-4o")
tokens = enc.encode("Your text here")
print(f"Token count: {len(tokens)}")
# Calculate cost
input_tokens = len(tokens)
cost = (input_tokens / 1_000_000) * 2.50 # GPT-4o input price
print(f"Input cost: ${cost:.6f}")Token Estimation Rules
- 1 token = approximately 4 characters in English
- 1 token = approximately 0.75 words
- 100 words = approximately 133 tokens
- Code typically uses more tokens than natural language
How to Reduce OpenAI API Costs
1. Choose the Right Model
Don't use GPT-5 for everything. Match model capability to task complexity:
- Simple tasks (classification, extraction): GPT-4o-mini ($0.15/1M)
- Standard tasks (chat, generation): GPT-4o ($2.50/1M)
- Complex reasoning: GPT-5 or o3 only when needed
2. Optimize Your Prompts
- Remove unnecessary whitespace and formatting
- Use concise system prompts, they're sent with every request
- Avoid repeating instructions in every message
3. Use Prompt Caching
OpenAI automatically caches prompt prefixes. Keep your system prompt consistent to get 50% off on cached tokens.
4. Batch Requests
Use the Batch API for non-time-sensitive workloads to get 50% discount on all tokens.
Frequently Asked Questions
How much does ChatGPT API cost per 1000 tokens?
ChatGPT API costs vary by model. GPT-4o costs $2.50 per 1M input tokens ($0.0025 per 1K) and $10 per 1M output tokens. GPT-4o-mini is much cheaper at $0.15 per 1M input tokens ($0.00015 per 1K). GPT-5 costs $15 per 1M input tokens. For the most cost-effective option, use GPT-4o-mini for simple tasks.
How many tokens can GPT-5 process?
GPT-5 has a context window of 256,000 tokens, allowing it to process approximately 192,000 words or 400+ pages of text in a single request. This is a significant upgrade from GPT-4's 128K context window.
What's the difference between input and output tokens?
Input tokens are the tokens in your prompt (what you send to the API), while output tokens are in the model's response. OpenAI charges different rates for each, output tokens are typically 3-4x more expensive because generating text requires more computation.
How do I count tokens in Python?
Use the official tiktoken library: import tiktoken; enc = tiktoken.encoding_for_model('gpt-4o'); tokens = enc.encode('your text'); print(len(tokens)). TokenCalc uses this same library for 100% accurate counts.