Price structure

The total price for annotating N items contains:

• Expert service cost: You set the price per task (with platform recommendation).

• Quality control (LLM QA): Fixed price per item, calculated automatically.

• Platform fee: Fixed price per item, calculated automatically.

How pricing works

Step 1: Select expert audience

You choose your target audience using filters:

• Languages: Select languages experts must know

• Specializations: Choose domain expertise if needed (medical, legal, technical, etc.)

  • General: Experts without specific domain expertise (for standard ML tasks)

How filters affect price

• More specific filters (rare languages, specialized domains) = higher price

• General audience = lower price

• More experts available = faster completion

Step 2: Select the number of items for launch

Step 3: Set task duration

You provide an estimate of how long it takes to complete one task:

• Based on your self check experience (see Self Check Process: Testing Your Configuration).

• Realistic time estimate (experts need adequate time).

• Too short = experts won’t take tasks.

Step 4: Set price per task

The agent calculates and recommends a price based on:

• Your task duration estimate.

• Selected expert audience parameters.

The recommendation includes

• Recommended price per task (with minimum limit).

• Expert availability: “500 experts available with your filters”.

• You can keep the recommended price.

• Or change it to a different amount.

• Platform fee & LLM QA are calculated automatically as a percentage of your price per task.

Step 5: Save settings

Step 6: Launch

1. Platform holds funds

   • Full amount is held at launch (including Expert Work, Quality Control, Platform Fee) (not charged yet)

   • Covers all items in your dataset

   If you don’t have enough funds in your account to start the launch, you need to top up your balance. You can do this by clicking Top Up next to Total Price or via the User Menu.

   For more details on topping up your balance, see the corresponding article Account Balance and Billing.

2. Deduction after delivery

   • After delivering results to you, cost per item is deducted from hold (including Expert Service Costs, Quality Control, Platform Fee).

   • You only pay for completed and delivered items Ready for Review.

3. If annotation process is cancelled

   • Money is returned for items that haven’t been taken into work by experts (Uploaded or Finding Expert status).

   • You only pay for work that was actually started.

Important

One Version = One Cost = One Launch

• Each project version has a fixed cost per task.

• Price is locked when you launch.

• Settings cannot be changed after launch.

• For different settings, create a new project version.