Vision & documents · AWS vs Google Cloud
Amazon Rekognition / Amazon Comprehend vs Cloud Vision API / Cloud Natural Language API / Model Armor
Rekognition and Comprehend moderation map to Cloud Vision SafeSearch for images, Natural Language moderateText for text, and Model Armor for AI prompts and responses.
Side-by-side comparison
| Aspect | Amazon Web Services | Google Cloud |
|---|---|---|
| Service | Amazon Rekognition Amazon Comprehend | Cloud Vision API Cloud Natural Language API Model Armor |
| Lifecycle | Streaming Events and Batch Image Content Moderation are in maintenance (Mar 31, 2026). Person Pathing was fully shut down Oct 31, 2025. Topic modeling, event detection, and prompt safety classification are no longer available to new customers (maintenance, Mar 31, 2026). Other features are unaffected. | Generally available Generally available Generally available |
| What it is | Prebuilt image and video analysis. Labels, text, moderation, face detection, comparison and search, celebrity recognition, and Face Liveness. NLP. Entities, key phrases, sentiment and targeted sentiment, language detection, PII, syntax, toxicity, and custom classification/entity recognition. | Pretrained image analysis (labels, objects, OCR, faces without identity, landmarks, logos, SafeSearch). Pretrained entity, sentiment, syntax, classification, and text moderation analysis. Screens prompts and responses for prompt injection, jailbreaks, harmful content, sensitive data, and malicious URLs, for any model. |
| Pricing model | Per image (Rekognition) or per text unit (Comprehend). | Cloud Vision is billed per 1,000 units. Natural Language is billed per 1,000-character unit. Model Armor is billed by tokens screened. |
| Docs | Amazon Rekognition docs Amazon Comprehend docs | Cloud Vision API docs Cloud Natural Language API docs Model Armor docs |
| Pricing page | Amazon Rekognition pricing Amazon Comprehend pricing | Cloud Vision API pricing Cloud Natural Language API pricing Model Armor pricing |
Key differences
- Moderation is spread across three products instead of two.
- SafeSearch returns likelihoods per category (adult, violence, racy, and more) rather than a moderation label taxonomy.
Pricing summary
Pricing models only. No prices are quoted because they change by region and over time.
- AWS
- Per image (Rekognition) or per text unit (Comprehend). Amazon Rekognition pricing → Amazon Comprehend pricing →
- GCP
- Cloud Vision is billed per 1,000 units. Natural Language is billed per 1,000-character unit. Model Armor is billed by tokens screened. Cloud Vision API pricing → Cloud Natural Language API pricing → Model Armor pricing →
And on Azure?
Direct match Azure's equivalent is Azure AI Content Safety. AWS vs Azure comparison →
FAQ
What is the Google Cloud equivalent of Amazon Rekognition and Amazon Comprehend?
Rekognition and Comprehend moderation map to Cloud Vision SafeSearch for images, Natural Language moderateText for text, and Model Armor for AI prompts and responses.
How does pricing compare between Amazon Rekognition and Amazon Comprehend and Cloud Vision API, Cloud Natural Language API and Model Armor?
AWS: Per image (Rekognition) or per text unit (Comprehend). Google Cloud: Cloud Vision is billed per 1,000 units. Natural Language is billed per 1,000-character unit. Model Armor is billed by tokens screened. Check the official pricing pages for current rates.
What is the Azure equivalent?
Azure AI Content Safety (direct match). Rekognition content moderation (and Comprehend toxicity detection for text) maps to Azure AI Content Safety.