Processing or generating 500 words (approximately 667 tokens) with Gemini 2.5 Pro ranges from $0.000208 (cached input) to $0.00667 (full generation).
You send 500 words as prompt context, documentation, or background knowledge.
Gemini 2.5 Pro drafts a complete 500 words article, chapter, or code module.
400 words input prompt + 100 words output reply.
| Model | Provider | Input Cost (500 words) | Cached Input Cost | Output Cost (500 words) | Context Limit |
|---|---|---|---|---|---|
| Gemini 2.5 Pro (Current) | $0.000834 | $0.000208 | $0.00667 | 1,000,000 | |
| GPT-5.6 Terra | openai | $0.00128 | $0.000128 | $0.00768 | 1,050,000 |
| GPT-5.4 Workhorse | openai | $0.0016 | $0.00016 | $0.0096 | 256,000 |
| o3-mini | openai | $0.000704 | $0.000352 | $0.002816 | 200,000 |
| o4-mini | openai | $0.000704 | $0.000176 | $0.002816 | 256,000 |
| o1-mini | openai | $0.000704 | $0.000352 | $0.002816 | 128,000 |
For Gemini 2.5 Pro (sentencepiece tokenizer), 500 words is approximately 667 tokens (an average ratio of 1.33 tokens per word in English). Code, technical vocabulary, and non-English scripts will have higher token densities.
Sending 500 words as input costs $0.000834 (or $0.000208 with prompt caching). Generating 500 words as output costs $0.00667. Output tokens are more expensive because autoregressive token generation requires significantly more computation per token.
Yes! If your input text is part of a repeated context, prompt caching saves 50% to 90% (costing $0.000208). For non-urgent asynchronous batch processing, 24-hour batch queues provide a flat 50% discount across all tokens (costing $0.000417 for input, $0.003335 for output).
At an average human reading speed of 250 words per minute, 500 words takes approximately 2 minutes to read. In contrast, Gemini 2.5 Pro can process or generate this text in seconds.