AI Providers

Configure inference engines, models (Claude, GPT-4o, Llama 3), and custom provider keys.

Overview

Asiyst is model-agnostic, allowing you to select the best foundational LLM for your speed, quality, and cost requirements.

Supported Models

Provider	Supported Models	Best For
Anthropic	Claude 3.5 Sonnet, Claude 3 Haiku	Complex reasoning, nuance, technical documentation
OpenAI	GPT-4o, GPT-4o-mini	High speed, multi-lingual performance, general FAQ
Meta / Groq	Llama 3.1 70B, Llama 3.1 8B	Ultra-low latency (< 150ms TTFT), cost optimization

Managed vs Bring Your Own Key (BYOK)

Managed Credits: Asiyst provides built-in model inference billed directly through your project credits.

Bring Your Own Key: Enter your Anthropic or OpenAI API keys in Project Settings -> AI Provider. Your keys are encrypted at rest using AES-256-GCM in Supabase Vault.

Fallback & Reliability

If a provider endpoint experiences transient downtime or rate limits (HTTP 429), the Asiyst inference router automatically fails over to your secondary configured model within 200ms.

See Models Reference for parameter configurations.