Precision vs Performance: Choosing the Right Rendering Mode

The Core Tradeoff

Sentience gives you a choice that traditional browsers don't: optimize for speed or optimize for accuracy. Understanding when to use each mode can save you thousands of dollars and dramatically improve your AI agent's performance.

Performance Engine

<500ms response time
2 credits per request
>95% accuracy

Precision Engine

~2s response time
10 credits per request
>99.5% accuracy

When to Use Performance Engine

Use the Performance Engine (default) for 95% of your requests. It's optimized for speed while maintaining excellent accuracy for most use cases.

Perfect For:

High-frequency automation: When you're processing thousands of pages per hour
Content extraction: RAG pipelines, data aggregation, content monitoring
AI agent training: When 95%+ accuracy is sufficient for your use case
Cost-sensitive workloads: When budget is a primary concern

Example: Training an AI Shopping Assistant

Imagine you're training an AI agent to help users find products online. You need to process thousands of product pages to build your training dataset:

1// Performance Engine is perfect for this
2{
3"url": "https://example-shop.com/product/12345",
4"mode": "map"  // Uses Performance Engine by default
5}
6
7// Response in ~400ms with 2 credits
8// Accurate enough for agent training

When to Use Precision Engine

Use the Precision Engine when accuracy is non-negotiable. It uses a reference browser engine for pixel-perfect rendering.

Perfect For:

Visual QA testing: Verifying pixel-perfect layouts across browsers
Ground truth dataset generation: When you need 100% accurate labels for ML training
Debugging layout issues: When Performance Engine results seem off
Legal/compliance validation: When you need defensible accuracy

Example: Visual Regression Testing

You're building a visual regression testing tool that needs to catch even 1-pixel differences:

1// Precision Engine for pixel-perfect accuracy
2{
3"url": "https://your-app.com/checkout",
4"mode": "map",
5"options": {
6  "render_quality": "precision"
7}
8}
9
10// Response in ~2s with 10 credits
11// Guaranteed pixel-perfect coordinates

The Hybrid Strategy (Best Practice)

The smartest teams use both engines strategically:

Recommended Workflow

1. Use Performance Engine for initial data collection
Process 10,000 pages with Performance Engine to build your initial dataset (20,000 credits)
2. Sample with Precision Engine for validation
Take 1% random sample (100 pages) and process with Precision Engine to verify accuracy (1,000 credits)
3. Use Precision Engine for edge cases
When Performance Engine returns low confidence scores, automatically retry with Precision Engine

Cost Comparison

Scenario	Performance	Precision	Hybrid
10,000 pages/day	20,000 credits $10/day	100,000 credits $50/day	21,000 credits $10.50/day
Response time	~400ms	~2s	~420ms avg
Accuracy	95-98%	99.5%+	99%+

The hybrid approach gives you 99%+ accuracy at just 5% higher cost than Performance-only!

Quick Decision Guide

Use Performance When:

Processing >1000 pages/hour
Speed is critical
Budget-conscious
95% accuracy is sufficient

Use Precision When:

Accuracy is non-negotiable
Visual QA testing
Dataset validation
Debugging issues

Try Both Engines

The best way to understand the difference is to try both. Start with Performance Engine and use Precision Engine when you need it:

1// Default: Performance Engine
2curl -X POST https://api.sentienceapi.com/v1/observe \
3-H "Authorization: Bearer YOUR_API_KEY" \
4-d '{"url": "https://example.com", "mode": "map"}'
5
6// Explicit: Precision Engine
7curl -X POST https://api.sentienceapi.com/v1/observe \
8-H "Authorization: Bearer YOUR_API_KEY" \
9-d '{
10  "url": "https://example.com",
11  "mode": "map",
12  "options": {"render_quality": "precision"}
13}'

Start Experimenting Today

Your free tier includes 200 credits. Try both engines and find the perfect balance for your use case.

Get Started Free