Processing or generating 1,000,000 words (approximately 1,333,000 tokens) with Perplexity Sonar Deep Research ranges from $0.6665 (cached input) to $10.664 (full generation).
Quick answer
For 1,000,000 words, Perplexity Sonar Deep Research processes approximately 1,333,000 tokens. That costs $2.666 as uncached input, $0.6665 with the modeled cache discount, or $10.664 when generated as output.
Method & trust
The estimate converts the selected English word count with the model tokenizer family, then applies the published input/output price per million tokens. Real prompts vary with code, formatting, language, and system instructions.
You send 1,000,000 words as prompt context, documentation, or background knowledge.
Perplexity Sonar Deep Research drafts a complete 1,000,000 words article, chapter, or code module.
800,000 words input prompt + 200,000 words output reply.
| Model | Provider | Input Cost (1,000,000 words) | Cached Input Cost | Output Cost (1,000,000 words) | Context Limit |
|---|---|---|---|---|---|
| Perplexity Sonar Deep Research (Current) | perplexity | $2.666 | $0.6665 | $10.664 | 128,000 |
| GPT-5.6 Sol | openai | $6.40 | $0.64 | $38.40 | 1,050,000 |
| GPT-5.6 Cyber | openai | $16.00 | $1.60 | $96.00 | 1,050,000 |
| GPT-5.5 Standard | openai | $6.40 | $0.64 | $38.40 | 512,000 |
| GPT-5.5 Pro | openai | $38.40 | $3.84 | $230.40 | 512,000 |
| o3-pro (Frontier Reasoning) | openai | $25.60 | $2.56 | $102.40 | 1,000,000 |
For Perplexity Sonar Deep Research (tiktoken_cl100k tokenizer), 1,000,000 words is approximately 1,333,000 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 1,000,000 words as input costs $2.666 (or $0.6665 with prompt caching). Generating 1,000,000 words as output costs $10.664. 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 $0.6665). For non-urgent asynchronous batch processing, 24-hour batch queues provide a flat 50% discount across all tokens (costing $1.333 for input, $5.332 for output).
At an average human reading speed of 250 words per minute, 1,000,000 words takes approximately 4000 minutes to read. In contrast, Perplexity Sonar Deep Research can process or generate this text in seconds.