Adola

About Adola

Adola is an OpenAI-compatible gateway for open-weight models, aimed at teams who want to run GPT-OSS, Qwen, Llama or DeepSeek without swapping out the client library they already ship. You keep your existing OpenAI client, change the model ID per request, and get streaming responses plus a per-request cost receipt back. The pitch that distinguishes it from a general router is explicitness: Adola runs the exact model you asked for or returns a clear error, never silently substituting a cheaper or differently-quantised variant when capacity is tight — which is the failure mode that makes multi-provider routers hard to benchmark against. Pricing is straightforwardly usage-based with no subscription, seat fee or monthly minimum; each of the four supported models publishes its own input and output rate and a per-successful-request minimum, so you know the cost of a call before your application makes it. The dashboard tracks tokens, latency and spend in one place, and project keys with usage controls let you cap a workload rather than discover the overrun on the invoice. The catalogue is small by design — four models with four clear prices rather than a hundred with variable availability — which is a real limitation if you need frontier closed models, and a real advantage if your problem is that a router's routing is the thing you cannot debug.

Key Features

Adola Pros & Cons

Tags