Models & Pricing

Choose the right model and OCR tier for your needs.

Parse requests accept a config object. Set parse tier with config.parse.model_tier ("basic", "standard", or "premium"). Matching tier can be set independently with config.matching.model_tier. OCR is configured with config.extract.type ("file" or "ocr").

Job matching is configured under config.matching and enabled by providing input.job_description. Matching adds credits equal to the selected matching tier.

When config.parse.skip: true, parse credits are not charged. Cost becomes extraction (if any) + matching.

Example (OCR enabled, Standard tier):

{
  "config": {
    "extract": { "type": "ocr" },
    "parse": { "model_tier": "standard" }
  }
}

Credits when the pipeline stops early: If text validation or the prompt check rejects a document before the LLM parse runs, you are still charged using the same formula: the model’s base credit cost plus any OCR credits if OCR ran. The LLM parse step is skipped; the charge is not reduced on that account. If matching fails after parse succeeds, parse credits are charged but matching credits are not.

Subscription Plans

Both plans include the same API surface and JSON schema. Differences are in monthly volume and document limits.

Credits reset at the start of each billing cycle. Unused credits do not roll over.

AI Models

Basic (1 credit)

• Best for: Standard resumes with clear formatting
• Accuracy: Good (95%+)
• Speed: Fast (~5-10 seconds)
• Cost: 1 credit per parse
• Availability: All plans

Standard (3 credits)

• Best for: Complex resumes, multiple formats
• Accuracy: Higher (97%+)
• Speed: Medium (~10-15 seconds)
• Cost: 3 credits per parse
• Availability: All plans

Premium (6 credits)

• Best for: Executive resumes, unusual formats
• Accuracy: Highest (98%+)
• Speed: Slower (~15-25 seconds)
• Cost: 6 credits per parse
• Availability: All plans

OCR Tiers

None (0 credits per page)

• Best for: Digital PDFs and clear native text
• Cost: 0 credits per page (text extraction only)
• API: Default when OCR is skipped (steps.ocr.skip: true)

Standard (+2 credits per page)

• Best for: Scanned documents, images with text
• Cost: +2 credits per page

Premium (+4 credits per page)

• Best for: Poor quality scans, complex layouts
• Cost: +4 credits per page

Pricing Examples

The examples below show credit cost per parse. Your monthly plan allowance covers these parses within your credit budget.

Digital PDF (2 pages)

• Basic model: 1 credit
• Standard model: 3 credits
• Premium model: 6 credits

Digital PDF (2 pages) + matching

• Basic model + matching: 2 credits
• Standard model + matching: 6 credits
• Premium model + matching: 12 credits

Scanned PDF (2 pages)

• Basic + Standard OCR: 5 credits
• Standard + Standard OCR: 7 credits
• Standard + Premium OCR: 11 credits

Scanned PDF (2 pages) + matching

• Basic + Standard OCR + matching: 6 credits
• Standard + Standard OCR + matching: 10 credits
• Standard + Premium OCR + matching: 14 credits

Scanned PDF (8 pages, Professional plan max)

• Basic + Standard OCR: 17 credits
• Standard + Premium OCR: 35 credits

How far does your plan go?

• Beginner (500 credits/mo): ~500 basic parses, or ~166 standard parses
• Professional (5,000 credits/mo): ~5,000 basic parses, or ~1,666 standard parses

Cost Optimization

Choose Model by Resume Type

• Simple, well-formatted: Basic
• Multiple jobs, complex: Standard
• Executive, unusual layout: Premium

Minimize OCR Costs

• Request digital PDFs when possible
• Leave OCR skipped (steps.ocr.skip: true, the default) when the file has extractable text
• Use Standard OCR for scanned documents
• Reserve Premium OCR for poor quality scans

Prompt check

Enabling the prompt check does not add extra credit cost beyond the normal model and OCR charges.

Matching

Enabling matching adds credits equal to the selected model tier when matching succeeds:

• Basic: +1
• Standard: +3
• Premium: +6

Monitor Usage

• Track average credits per parse in dashboard
• Identify patterns (model selection, file types)
• Adjust strategy based on actual usage