---
title: Providers Overview
description: BYOK multi-provider routing and architecture in NIKI.
sidebar:
  order: 1
  label: Providers Overview
---

## BYOK & Multi-Provider Architecture

NIKI is built with a **Bring Your Own Keys (BYOK)** philosophy. You are never locked into a single model vendor or proprietary proxy.

Each agent role in the pipeline can be mapped to an entirely different provider and model:

```toml
# Mix providers freely across roles
[agents.planner]
provider = "anthropic"
model    = "claude-sonnet-4-20250514"

[agents.coder]
provider = "anthropic"
model    = "claude-sonnet-4-20250514"

[agents.tester]
provider = "openai"
model    = "gpt-4o-mini"

[agents.reviewer]
provider = "anthropic"
model    = "claude-sonnet-4-20250514"
```

---

## Supported Providers

* **Anthropic:** Claude 3.5 Sonnet, Claude 3.7 Sonnet, Claude 3.5 Haiku.
* **OpenAI:** GPT-4o, GPT-4o-mini, o1, o3-mini.
* **OpenRouter:** Universal gateway accessing hundreds of models with unified billing.
* **NVIDIA NIM & Together AI:** Fast inference on open-weights models (DeepSeek-R1, Llama-3.3-70B, Qwen-2.5-Coder).
* **Groq:** Ultra-low latency inference for quick Tester and single-agent passes.
* **OpenCode Zen:** One-key gateway (MiniMax, GLM, Kimi models, free tiers) — see [AI Gateways](./ai-gateways).
* **Kimi Code:** Moonshot's coding gateway (`k3-256k`) — see [AI Gateways](./ai-gateways).
* **KiloCode Gateway:** Hundreds of models as `vendor/model` behind one key — see [AI Gateways](./ai-gateways).
* **Google Gemini:** Gemini 1.5 Pro, Gemini 2.0 Flash.
* **Ollama:** 100% offline, local GPU inference.