Glossary

The base URL used to access the AlphaEdge API. All requests must be sent to https://api-endpoints.alphaedge-ai.com followed by the specific endpoint path (e.g. /models).

Unique authentication key required to access the AlphaEdge API. It must be included in the X-API-Key header of each request in the form TA_CLE. You can obtain an API key from your dashboard.

The AlphaEdge API uses REST (Representational State Transfer) architecture, a web architecture style that uses standard HTTP methods (GET, POST, PUT, DELETE) to communicate with the server.

Mode de réponse où les données sont envoyées progressivement au client au fur et à mesure de leur génération, plutôt qu'en une seule réponse complète.

This feature is not yet exposed by the public gateway. Current endpoints (transcription, OCR) return a complete response.

An artificial intelligence model trained to understand and generate natural text. AlphaEdge models are optimized for speed and energy efficiency while maintaining high performance.

The maximum context length the model can process in a single request, measured in tokens. AlphaEdge models support up to 128K tokens of context, allowing very long documents to be processed.

The maximum length of the response generated by the model, measured in tokens. By default, AlphaEdge models generate up to 4K tokens, with a maximum of 8K tokens for longer responses.

A model capable of processing multiple types of input data, such as text and images. AlphaEdge Vision is an example of a multimodal model that can analyze images and generate text descriptions.

The smallest unit of text that the model recognizes. A token can be a full word, part of a word, a number, or even a punctuation mark. Tokenization varies by language and context.

The tokens you send to the API in your request, including system messages, user messages, and any other input content. Input tokens are billed at the input rate.

The tokens generated by the model in its response. Output tokens are billed at the output rate, which is generally higher than the input rate because generation requires more resources.

The process of splitting text into tokens. Each model uses a specific tokenizer that determines how text is segmented. Tokenization affects the total number of tokens and therefore the cost of your request.

Endpoint conversationnel permettant d'envoyer une série de messages (système, utilisateur, assistant) et de recevoir une réponse contextuelle basée sur l'historique.

Not available on the public gateway today. Current endpoints cover audio transcription and OCR.

Fonctionnalité permettant au modèle d'appeler des fonctions externes pendant la génération, pour intégrer des systèmes tiers.

Not available on the public gateway today.

Capacité du modèle à générer des réponses dans un format structuré spécifique (JSON, XML, etc.) selon un schéma défini.

Not available on the public gateway today.

Technique permettant au modèle de compléter du code ou du texte en remplissant une partie manquante au milieu d'un contexte.

Not available on the public gateway today.

Fonctionnalité permettant de fournir un préfixe de texte que le modèle doit compléter, utile pour l'autocomplétion.

Not available on the public gateway today.

Mécanisme permettant au modèle d'utiliser des outils externes (fonctions, APIs, bases de données) pendant la génération.

Not available on the public gateway today.

AlphaEdge pricing is expressed in units per million (1M) tokens. The total cost of a request is calculated by multiplying the number of tokens used by the corresponding price, then dividing by 1,000,000.

Credits you have purchased and added to your account. Charges are deducted from this balance first after the granted balance is exhausted.

Free or promotional credits granted by AlphaEdge. This balance is used first before the topped-up balance to cover API usage charges.