Available Models
Choose the right model for your alignment tasks. Higher-credit models generally produce more accurate results.
OpenAI's newest and most capable model. Best accuracy for complex layouts, dense text, and difficult scans. OCR only.
OpenAI's workhorse model. Strong accuracy at a fraction of the flagship's cost — a solid default for most documents. OCR only.
OpenAI's budget-friendly model. Fast and inexpensive — good for straightforward documents with clear formatting. OCR only.
Anthropic's most capable model. Top accuracy for complex layouts and nuanced translations. PDF native.
Anthropic's balanced model. Excellent at following alignment instructions precisely. PDF native.
Anthropic's fastest model. Great speed-to-quality ratio for high-volume work. PDF native.
Google's most cost-efficient multimodal model. Fast, good for straightforward documents and multilingual content. PDF native.
Balanced Google multimodal model. Stronger reasoning and OCR than Flash-Lite for moderately complex layouts. PDF native.
Google's flagship multimodal model. Highest accuracy for complex layouts, dense text, and difficult scans. Preview release. PDF native.
1 credit models
Fast, affordable models ideal for straightforward documents with clear formatting and standard layouts.
2 credit models
Higher-accuracy models well suited to complex layouts, dense text, or moderate-quality scans.
3 credit models
Top-tier models with the strongest reasoning and OCR. Best for the most challenging documents — poor contrast, handwriting, or intricate multi-column layouts.
Credit requirements correspond to one PDF page (source language and target language PDFs) processed by the selected model. Batch processing applies a 50% discount.
Monthly plan credits reset at the start of each billing cycle — unused credits do not roll over. Add-on credits are kept even if you change or cancel your plan.