Adola

Adola is an API service for running open models through a single OpenAI-compatible endpoint. It is built for teams that want to ship against GPT-OSS, Qwen, Llama, and DeepSeek without rebuilding an existing stack.

The service supports changing models with one parameter, streaming responses as they arrive, and checking request cost before building. It also supports moving between reasoning, coding, multilingual, and general-purpose models by changing a single parameter. The site says existing OpenAI-compatible request code can stay the same, and that migration involves changing the base URL and API key while keeping the client, message format, streaming flow, and response parsing already in use. It also notes that the response shape remains the one the application already handles.

Adola exposes its service at a base URL and provides an API key for access. The quickstart example shows an OpenAI client pointed at https://api.adola.app/v1, and the page describes OpenAI-compatible server-sent events for streaming. It also mentions a live demo, product docs, models and pricing information, and resources. Pricing details are presented as model pricing, with request-level cost visibility, and the first 100 API requests each month are free with no card and no subscription required. The page also says users can create a key, make a real model call, and decide whether Adola belongs in their stack before adding a card.