Airtop uses a variety of LLMs from OpenAI, Anthropic, Google, and custom fine-tuned models. Airtop may dynamically switch between different models based on the workload and model capabilities. Additionally, Airtop may failover to a different model if a model is experiencing high latency or downtime. Airtop additionally uses vision capable models to provide visual data to aid in the identification of content, making prompting against sites more accurate.
Airtop uses a credit system to track all types of usage. If you use an AI API to, say, query a page, Airtop will charge your account for the cost of the API call in the form of credits. Airtop also charges credits for CPU time of the browser instance, as well as proxy bandwidth, if you are using a proxy.
Because Airtop uses a variety of LLMs, sometimes within the same API call, Airtop cannot charge for token usage directly since different token consumption may have a different cost. Instead, Airtop assigns a credit value to each LLM’s token use and presents a single unified cost to developers. We are constantly working to lower the cost of LLM usage and pass on savings to developers.
No. In the cases where Airtop uses an LLM hosted by a third party, Airtop always does so under contracts that prohibit the third party from using Airtop’s customers’ data for their own model training.
Yes, Airtop is SOC 2 compliant. If you need access to our SOC 2 report, please reach out to us at support@airtop.ai.
Yes, Airtop is HIPAA compliant. If you need access to our HIPAA report, please reach out to us at support@airtop.ai.
Airtop is designed to be used for good things. We do not allow any use of Airtop to engage in any activity that violates laws or the terms of service of the websites you are visiting. Please be aware that the excessive or abusive use of bots, or automation tools on some websites can lead to penalties, including temporary or even permanent restrictions on your account. We recommend you read and adhere to the specific terms of the websites you are visiting and using to avoid any disruptions or issues. We do not assume any responsibility for the consequences of abuse.