Which LLM Providers Are Supported

The G8KEPR AI Gateway routes your inference traffic across 14 built-in LLM providers, plus a Custom option for your own endpoints. Every request passes through the same security, routing, and cost-tracking layer no matter which provider serves it, so you can mix and match without changing your app code.

How provider support works

Each provider ships with a first-party adapter that handles authentication, message formatting, cost tracking, response streaming, and error normalization for you. You point your traffic at the Gateway once, and it translates to each provider's native API format behind the scenes.

Supported providers

• ▸Anthropic — Claude models, including the Claude 3 and Claude 4 families. Streaming supported.
• ▸OpenAI — GPT-4 and later models, plus the o-series reasoning models. Streaming supported.
• ▸Google — Gemini models via the Gemini API. Streaming supported.
• ▸Azure OpenAI — OpenAI models via Azure, with bring-your-own-key and regional endpoints.
• ▸AWS Bedrock — Claude, Titan, and Llama models via Bedrock.
• ▸Groq — Llama and Mixtral models.
• ▸Cohere — Command and Embed models.
• ▸Mistral — Mistral models, with streaming.
• ▸DeepSeek — DeepSeek Chat and Coder models.
• ▸Fireworks AI — custom model IDs on Fireworks.
• ▸Together AI — Llama and Mistral models.
• ▸xAI — Grok models.
• ▸Cerebras — models served on Cerebras.
• ▸Ollama — self-hosted models for local deployment.
• ▸Custom (BYOI) — your own user-defined endpoints, with API-key authentication.

Add a provider

1. 1.Open AI Gateway → Providers in your dashboard.
2. 2.Choose a provider and enter its API key. Keys are stored encrypted at rest.
3. 3.Set an optional load-balancing weight and mark the provider active.
4. 4.Save. The Gateway validates the provider and begins routing eligible traffic to it.

Need a provider that isn't listed? Reach out through Contact Support and we'll help you wire up a Custom integration.