Models Overview

SovereignEG serves models through an OpenAI-compatible API — switch chat models by changing one string. See the Model Library for what is live today.

Available models

Token models are priced in EGP per 1M tokens and image models per image, charged to your prepaid balance. Some models also bill cached, audio or image input at their own rates — see pricing in GET /v1/models or the model's page in the dashboard. Live means the model has a configured route, not a real-time health check. Coming soon models are listed but cannot be called yet.

Choosing a model

Chat models share one OpenAI-compatible endpoint, so you switch between them by changing the model string. Embedding models use /v1/embeddings, image models /v1/images/generations and decision models /v1/decisions (OpenAI's Decisions shape) or /v1/systemone (TypeSafe's System One shape). When picking a chat model:

• Latency + cost — smaller models return tokens faster and cost less per token. A good default for chat, drafting, and high-volume classification.
• Quality — larger models handle complex reasoning, long instructions, and code generation better. Reach for them when a smaller model falls short.
• Arabic — multilingual models trained with Arabic data (for example Qwen-class) handle Arabic well; test two or three on your own prompts. See the Arabic Guide for recommendations.
• Long context — sort by context window in the Model Library when you need to fit large documents into a single request.

Live status, context windows, and per-model EGP pricing always reflect the current catalog on the Model Library.