Vision AI Pricing REST API & WEB UI
All accounts are provided with a small amount of vision compute to perform basic experimentation online or perform API testing. To continue using Vision AI, more compute will be needed. This can be managed from within the dashboard or account page.
Vision AI pricing behaves differently than other more easily measured forms of artificial intelligence.
This is because there are a large number of different solutions that can be achieved with computer vision.
To provide a simplified integration, astica offers a single API endpoint that enables access to several vision AI models within a single API call.
The cost of using vision AI will depend on the parameters used within your API call.
- asticaVision 2.1 allows you to access v1.0_full capabilities in a single API call. v1.0 pricing applies, however v2.0 functionality is prioritized.
- GPT-S based functionality require data points of other parameters within the same API call. Increase accuracy and output quality of GPT by including all relevant parameters.
- All compute usage and costs of use will be calculated and displayed online. View Usage.
-
Base64 Input Ingress Cost:
The API now supports base64 image inputs, allowing inference on local image files.
If you choose to use this method, an additional ingress/egress charge (1mb min/request) will automatically be added to the API call ($0.0005/mb)
While it can impact Vision AI output, resizing large images is often recommended.
Using https url image inputs will avoid this cost, and also offer a faster response time.
asticaVision v1.0_full offers a variety of computer vision solutions at a low and uniform cost.
asticaVision v2.0_full enables access to more powerful computer vision capabilities at a slightly increased cost.
asticaVision includes a simplified manner of using GPT-S to describe an image.
This provides a detailed description of an image consisting of multiple sentences, forming a coherent paragraph(s) that describes the image contents.
GPT-S description can be generated with both asticaVision 1.0_full and 2.0_full.
The GPT-S capability is defined by the number of other parameters used. The GPT-S description is unlikely to comment on facial features unless the face detection parameter has been used. Providing all parameters for any GPT-S requests will offer the highest quality output possible.