This repository provides a definitive, community-curated list of Large Language Model (LLM) inference APIs that offer free tiers or substantial trial periods. It's designed for developers, researchers, and startups to experiment and build prototypes without initial financial commitment, while promoting responsible API usage.
Each entry is validated and includes critical details on rate limits, model availability, and usage policies.
The repository is structured to maintain data integrity and facilitate automated validation. All data is stored in a structured format, and Python scripts are used to enforce schema compliance and check for dead links.
sh . ├── .github/ # GitHub Actions, issue templates, and community health files ├── data/ │ └── providers.json # Single source of truth for all API provider data ├── scripts/ │ ├── init.py │ └── validate.py # Python script to validate the schema and content of providers.json ├── .gitignore ├── AGENTS.md # Directives for AI development agents ├── LICENSE ├── pyproject.toml # Project metadata and dependencies (uv/Ruff) └── README.md # This file
- 💡 Value Proposition
- 📂 Project Architecture
- 🤖 AI Agent Directives
- 🚀 Getting Started
- 💻 Available Scripts
- API Provider List
- 🤝 Contributing
- ⚖️ License
🤖 AI Agent Directives (System Instructions)
Role: You are a Senior Principal Software Architect and Master Technical Copywriter with 40+ years of elite industry experience. You operate with absolute precision, enforcing FAANG-level standards. Context: Current Date is December 2025. You are building for the 2026 standard. Output Standard: Deliver EXECUTION-ONLY results. Philosophy: "Zero-Defect, High-Velocity, Future-Proof."
This repository, LLM-Inference-API-Resource-Aggregator, is a data-centric project with Python-based validation scripts.
- PRIMARY SCENARIO: DATA VALIDATION & AGGREGATION (Python)
- Stack: This project leverages Python 3.11+. Key tools include uv (for package management), Ruff (for ultra-fast linting and formatting), and Pytest (for robust unit and integration testing of validation logic).
- Architecture: The core of this project is the structured data file
data/providers.json. All Python scripts in thescripts/directory must adhere to a Modular Monolith pattern. Each script should have a single responsibility (e.g., schema validation, link checking) and be executable independently. - Data Integrity: The highest priority is data accuracy. All contributions that modify
data/providers.jsonmust pass the validation workflow (uv run validate) before being considered for merge.
- Setup Environment:
uv venv && uv sync - Run Linter/Formatter:
uv run lint - Run Validation Checks:
uv run validate - Run All Tests:
uv run test
To get started with local development and run validation scripts, follow these steps. This project uses uv for package management.
-
Clone the repository: sh git clone https://github.com/chirag127/LLM-Inference-API-Resource-Aggregator.git cd LLM-Inference-API-Resource-Aggregator
-
Create a virtual environment and install dependencies: sh
pip install uv
uv venv
uv sync
All scripts are managed via uv and defined in pyproject.toml.
| Command | Description |
|---|---|
uv run validate |
Validates the schema and contents of data/providers.json. |
uv run lint |
Runs the Ruff linter and formatter to check code quality. |
uv run test |
Executes the Pytest test suite for validation logic. |
Below is a curated list of providers. For the complete, machine-readable list, please see data/providers.json.
| Provider | Key Models | Free Tier / Trial Details | Rate Limits (Free Tier) | Notes |
|---|---|---|---|---|
| OpenAI | gpt-4o-mini, gpt-3.5-turbo |
$5 in free credits for new API users (expires after 3 months). | Varies by model; check official docs. | The original industry standard. Credits are a one-time offer. |
| Google AI | gemini-1.5-flash |
Free Tier: 2 RPM, 32K TPM, 50 RPD. | 2 RPM (Requests/Min), 50 RPD (Requests/Day) | Excellent free tier for low-traffic applications. |
| Anthropic | claude-3-haiku |
$5 in free credits for new API users. | Varies; check official docs. | Known for strong performance on reasoning and safety. |
| Groq | llama3-8b, gemma-7b |
Free access during beta/promotional periods. | Generous but subject to change. | Focus on extremely high-speed inference (tokens/sec). |
| Together AI | llama3-8b, mixtral-8x7b |
$25 in free credits for new users. | Varies by model. | Offers a wide variety of open-source models via a unified API. |
| Provider | Key Models | Free Tier / Trial Details | Rate Limits (Free Tier) | Notes |
|---|---|---|---|---|
| OpenAI | text-embedding-3-small |
Included in the initial $5 free credit. | Varies. | High-performance embedding models. |
| Google AI | text-embedding-004 |
Free Tier: 750 RPM, 100K TPM, 100K RPD. | 750 RPM (Requests/Min) | A very generous free tier for embedding-heavy tasks. |
This list is actively maintained. Please consider contributing if you find an inaccuracy or a new provider.
Contributions are welcome and essential to keeping this resource up-to-date! Please read the Contributing Guidelines for details on our code of conduct and the process for submitting pull requests.
This project is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License - see the LICENSE file for details.