What methodology does the Scope3 API use to model and measure the emissions of AI products?
The Scope3 AI API implements an open source modeling methodology. Pull requests and issues are appreciated and may be provided via GitHub.How far back does Scope3’s data go?
The AI API is currently in active development and historical data is not consistent with methodology changes. When the methodology and API are ready for production use, we will start to publish monthly updates and provide historical consistency.Are there limits to the number or frequency of API calls that can be made?
There are no hard limits, and the API was designed with scale in mind. That said, it is not designed to be a real-time system.What is rate limiting policy?
The Scope3 API does not have a strict rate-limiting policy; instead, the API infrastructure will scale to meet the demand. During high load, the API may return a “429 Too Many Requests” HTTP response code. If your application receives a 429 response, it should retry the request after a small delay.How many rows can I send in a single request?
For best performance, we suggest sending 4,000 to 8,000 rows per request. The API will accept and process more than this, but your overall throughput will be higher if you split your workload into the suggested batch sizes above.Should I cache results?
In general, we do not recommend caching. If you are looking for ways to reduce the number of requests, please reach out to Scope3 support for best practices based on your use case.What countries are supported?
There are four dimensions to how well we support a country:- Do we have emissions intensity data for the electricity grid
- Do we have an accurate model of telco and consumer device emissions
- Do we have contributed data from key ad tech vendors in the market
- Do we model the idiosyncrasies of key publishers in the market