GCP Vision AI cost: per image, by feature
Cloud Vision bills per image per feature (label, face, text, object, safe-search, and more), so requesting several features on one image bills for each, with a free tier and volume discounts. Analyzing more features than needed drives the bill.
Quick answer
Cloud Vision bills per image per feature (label detection, face detection, text/OCR, object localization, safe-search, and others), each priced separately, so requesting several features on one image bills for each, with a monthly free tier and tiered volume discounts. Cost scales with images times features, so requesting only the features you need and caching results are the levers.
Google Cloud Vision analyzes images for labels, faces, text, objects, and more. It is priced per image per feature, so the unit is not the image but the image-feature pair. Requesting several features on one image bills for each, which is where cost grows if you request more analysis than the use case needs.
Per image, per feature
| Feature (examples) | Billing |
|---|---|
| Label detection | Per image |
| Text detection / OCR | Per image |
| Face / object / safe-search | Per image, each feature separate |
Each feature is billed per image, at tiered rates that fall with volume, with a monthly free tier. So an image analyzed for labels, text, and objects bills three times, once per feature. Requesting only the features the use case needs avoids paying for analysis you do not use.
Controlling Vision AI cost
Request only the features each use case needs (do not run every feature on every image), cache results so you do not re-analyze the same image, pre-filter to skip irrelevant images before calling the API, and use the free tier for development. Because billing is per image per feature, the levers are feature selection and caching, exactly as with Rekognition and Azure AI Services.
FAQ
How is GCP Cloud Vision priced?
Per image per feature (label detection, text/OCR, face detection, object localization, safe-search, and others), each priced separately, so requesting several features on one image bills for each. Rates are tiered and fall with volume, with a monthly free tier. Cost scales with images times features requested.
How do I reduce Cloud Vision cost?
Request only the features each use case needs rather than running every feature on every image, cache results to avoid re-analyzing the same image, pre-filter to skip irrelevant images before calling the API, and use the free tier for development. Feature selection and caching are the main levers.
Does requesting multiple Vision features cost more?
Yes. Each feature, label detection, text detection, face detection, object localization, is billed per image separately, so an image analyzed for three features bills three times. Requesting only the features the use case requires, rather than all of them, avoids multiplying the per-image cost.
Does Cloud Vision have a free tier?
Yes, a monthly free allowance per feature that covers low-volume usage and testing. Beyond it you pay per image per feature at tiered rates that fall with volume. The free tier makes low-volume experimentation essentially free across the features.
What drives Cloud Vision cost?
The number of images times the number of features requested per image. High image volume and requesting many features per image both multiply cost. Re-analyzing the same images without caching also wastes calls. Requesting only needed features and caching results keeps cost proportional to real need.
Does C3X estimate Cloud Vision cost?
Cloud Vision cost is driven by images and features analyzed, which are usage inputs. C3X prices the surrounding infrastructure, and you model your image volume and the features each use case needs to estimate the per-image-per-feature charges.
What to do next
Estimate the infrastructure around your vision workload before you build it. C3X reads your Terraform and prices your resources against a live catalog. Start with the quickstart.
Related posts
- GCP Document AI cost: per page, by processor type
Document AI bills per page processed, with rates by processor, cheap general OCR, more for form and specialized parsers, and higher for custom and generative processors. Running specialized processors on simple documents over-pays. Here is the model.
- Vertex AI cost optimization guide: controlling GCP ML spend
Vertex AI cost spans training jobs, online prediction endpoints, notebooks, and pipelines, and the biggest drains are always-on endpoints and idle notebooks. Here is how to optimize each part of GCP's managed ML platform.
- GCP Speech-to-Text cost: per minute, by model and features
Speech-to-Text bills per minute of audio (rounded up), with higher rates for premium and specialized models and features like diarization. Long audio and premium models drive the bill. Here is the per-minute model and how to control it.
- Vertex AI vs GKE for GPU workloads: comparing the two GCP paths
On Google Cloud you can run models on Vertex AI or on your own GKE cluster with GPU node pools. The hardware rates are similar, so the decision comes down to cluster fees, discount eligibility, and who operates it.
Share this post
Try C3X on your own Terraform
Free and open source. No API key required. One command to install, one command to estimate.