Gorilla

Free

An API store for LLMs ![GitHub Repo stars](https://img.shields.io/github/stars/ShishirPatil/gorilla?style=social)

FreeFree tier

Type

Open Source

Visit Website GitHub (40 stars)

About Gorilla

Gorilla is an open-source framework from UC Berkeley that enables large language models (LLMs) to use tools by invoking APIs. Given a natural language query, Gorilla generates semantically and syntactically correct API calls. The project includes the Berkeley Function Calling Leaderboard (BFCL) for evaluating function-calling capabilities, and has released several models including OpenFunctions, OpenFunctions v2, and gorilla-cli. Gorilla supports multiple API hubs like Torch Hub and TensorFlow Hub, and its commercially usable models are licensed under Apache 2.0. The project also introduces RAFT for domain-specific RAG and the Agent Arena for comparing agentic systems.

Key Features

API invocation from natural language queries

Training and evaluation of LLMs for function calls

Berkeley Function Calling Leaderboard (BFCL) for objective benchmarking

Commercially usable Apache 2.0 licensed models

OpenFunctions and OpenFunctions v2 models for state-of-the-art function calling

gorilla-cli for command-line interface interactions

Support for multiple API stores (Torch Hub, TensorFlow Hub, etc.)

RAFT for domain-specific retrieval-augmented generation

Agent Arena for comparing agent performance in tasks like search and finance

Pros & Cons

Pros

Open source with permissive Apache 2.0 license
State-of-the-art function calling capabilities for open-source LLMs
Active community and frequent updates with new benchmarks and features
Includes rigorous leaderboards for objective evaluation
Supports multiple API platforms and is extensible

Best For

Tool use and API discovery for LLMsBuilding autonomous agents that invoke external servicesFunction calling for enterprise workflows and automationRetrieval-augmented generation (RAG) in domain-specific contextsMulti-step reasoning and error recovery in agentic systems

FAQ

What is Gorilla?

Gorilla is a large language model (LLM) framework that enables LLMs to use tools by invoking APIs. Given a natural language query, it generates semantically and syntactically correct API calls.

What is the license for Gorilla models?

Gorilla models are released under the Apache 2.0 license, making them commercially usable.

What is the Berkeley Function Calling Leaderboard?

The Berkeley Function Calling Leaderboard (BFCL) is a benchmark for evaluating the function-calling capabilities of LLMs. It includes categories for multi-turn, multi-step, and agentic scenarios.