This project curates a collection of free Large Language Model (LLM) APIs, targeting developers and researchers seeking accessible AI model endpoints. It centralizes discovery and utilization of various LLM services, prioritizing APIs with rate limits over token caps and offering a diverse range of popular and newer models.

How It Works

The collection aggregates publicly available LLM APIs adhering to an OpenAI-compatible format. Selection prioritizes APIs limiting request rates (RPM/RPD) over token usage for broader accessibility. It aims to include diverse sources featuring newer, high-performing models with sufficient rate limits for practical use.

Quick Start & Requirements

This repository is a curated list of external API endpoints, not a deployable application. Users must register with individual API providers, obtain keys, and integrate them. Specific requirements vary by provider.

Highlighted Details

• SiliconFlow: Offers 1000 RPM per model, including DeepSeek-R1-0528, Qwen3-8B, and GLM-4-9B.
• OpenRouter: Provides 20 RPM / 200 RPD per model, featuring models like qwen/qwen3-coder:free and nvidia/nemotron-3-nano-30b.
• Google Gemini: Accessible via generativelanguage.googleapis.com/v1beta/openai with varying rate limits (e.g., 5 RPM / 20 RPD for gemini-3-flash-preview).
• NVIDIA NIM: Supports 40 RPM, including models like deepseek-ai/deepseek-v4-flash and nvidia/nemotron-3-ultra-550b.
• HuggingFace API: Offers 300 RPH (Hour) with models such as deepseek-ai/DeepSeek-V4-Pro and google/gemma-4-31B-it.
• Groq API: Features 30 RPM / 1000 RPD, including openai/gpt-oss-120b and qwen/qwen3-32b.
• Other Providers: Numerous services including Intern AI, Bigmodel, Github Models, Panda API, LLM7, ModelScope, Kilo Gateway, Celebras, Mistral, LongCat, OpenCode Zen, DXNT, and Agens AI, each with distinct limits and models.

Maintenance & Community

Maintained by a single individual focused on diligent updates. Contributions and sharing of additional API sources are welcomed.

Licensing & Compatibility

No software license is specified in the README. Users must consult individual API provider terms of service. Commercial use compatibility depends on each external API's licensing.

Limitations & Caveats

This is a curated list, not a deployable application, requiring users to manage individual API integrations and keys. Rate limits and model availability are subject to change by providers. Maintenance relies on a single contributor, potentially affecting update speed.