@tanstack/ai-llmgateway weekly downloads statistics

@tanstack/ai-llmgateway

LLM Gateway adapter for TanStack AI — one OpenAI-compatible endpoint that routes chat, tool calling, and structured outputs to hundreds of models across many providers.

Installation

Setup

Get your API key from the LLM Gateway dashboard and set it as an environment variable:

Usage

Text/Chat Adapter

With Explicit API Key

Self-Hosted Gateways

LLM Gateway is open source and self-hostable — point baseURL at your own deployment:

Models

Any model listed on llmgateway.io/models works — pass its id as the model name. A curated set of flagship models additionally carries per-model type metadata (input modalities, provider options) with autocomplete, including gpt-5.6-terra, claude-sonnet-5, gemini-pro-latest, kimi-k3, glm-5.2, deepseek-v4-pro, and more (see LLMGATEWAY_CHAT_MODELS).

Model ids accept an optional provider/ prefix to pin routing to a specific provider:

Features

• ✅ Streaming chat completions • ✅ Structured output (JSON Schema) • ✅ Function/tool calling • ✅ Multimodal input (text + images for vision models) • ✅ Reasoning output (reasoning_content deltas from reasoning models) • ✅ Summarization (llmGatewaySummarize)

Tree-Shakeable Adapters

This package uses tree-shakeable adapters, so you only import what you need:

Documentation

• TanStack AI Documentation • LLM Gateway Documentation

Description from the npm listing.