Processing or generating 1,000 words (approximately 1,333 tokens) with Perplexity Sonar ranges from $0.000333 (cached input) to $0.001333 (full generation).
You send 1,000 words as prompt context, documentation, or background knowledge.
Perplexity Sonar drafts a complete 1,000 words article, chapter, or code module.
800 words input prompt + 200 words output reply.
| Model | Provider | Input Cost (1,000 words) | Cached Input Cost | Output Cost (1,000 words) | Context Limit |
|---|---|---|---|---|---|
| Perplexity Sonar (Current) | perplexity | $0.001333 | $0.000333 | $0.001333 | 128,000 |
| GPT-5.6 Luna | openai | $0.000256 | $0.000026 | $0.001536 | 1,050,000 |
| GPT-5.4 mini | openai | $0.00096 | $0.000096 | $0.00576 | 256,000 |
| GPT-5.4 nano | openai | $0.000256 | $0.000026 | $0.0016 | 128,000 |
| GPT-4o mini | openai | $0.000192 | $0.000096 | $0.000768 | 128,000 |
| Claude Haiku 4.5 | anthropic | $0.00132 | $0.000132 | $0.0066 | 1,000,000 |
For Perplexity Sonar (tiktoken_cl100k tokenizer), 1,000 words is approximately 1,333 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 1,000 words as input costs $0.001333 (or $0.000333 with prompt caching). Generating 1,000 words as output costs $0.001333. 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 $0.000333). For non-urgent asynchronous batch processing, 24-hour batch queues provide a flat 50% discount across all tokens (costing $0.000667 for input, $0.000667 for output).
At an average human reading speed of 250 words per minute, 1,000 words takes approximately 4 minutes to read. In contrast, Perplexity Sonar can process or generate this text in seconds.