A standard 10 pages document (~5,000 words / 6,665 tokens) costs $0.000333 to ingest and $0.000373 to generate a comprehensive executive summary using Llama 3.1 8B Instruct.
Feed the entire 10 pages document into context to extract fields, entities, or answer queries.
Ingest 10 pages and generate a structured 500-token executive summary with key takeaways.
Translate or reformat the entire 10 pages document into an equivalent length output.
| Model | Provider | Ingestion (10 pages) | Cached Ingestion | Summary Cost | Context Limit |
|---|---|---|---|---|---|
| Llama 3.1 8B Instruct (Current) | meta | $0.000333 | $0.000083 | $0.000373 | 128,000 |
| GPT-5.6 Luna | openai | $0.00128 | $0.000128 | $0.00188 | 1,050,000 |
| GPT-5.4 mini | openai | $0.0048 | $0.00048 | $0.00705 | 256,000 |
| GPT-5.4 nano | openai | $0.00128 | $0.000128 | $0.001905 | 128,000 |
| GPT-4o mini | openai | $0.00096 | $0.00048 | $0.00126 | 128,000 |
| Claude Haiku 4.5 | anthropic | $0.0066 | $0.00066 | $0.0091 | 1,000,000 |
Assuming a standard single-spaced document with ~500 words per page, a 10 pages document contains approximately 5,000 words, which translates to roughly 6,665 tokens using Llama 3.1 8B Instruct's tokenizer. Dense PDF tables, legal boilerplate, and OCR scans may increase this by 20–30%.
Llama 3.1 8B Instruct has a context window of 128,000 tokens. A 10 pages document consumes only 5.21% of its total window, easily fitting in a single prompt without requiring chunking or vector search.
Ingesting the document and outputting a concise 500-token executive summary costs $0.000373. Using 24-hour async batch API queues drops this cost to $0.000187.
If your workflow repeatedly queries or chats with this same 10 pages document, prompt caching reduces subsequent turn input costs by up to 90% (to $0.000083 per turn instead of $0.000333).