Browse All AI Tools

AI data labeling and RLHF for enterprise AI teams

Scale AI is the leading data platform for AI development, providing high-quality training data labeling, RLHF (Reinforcement Learning from Human Feedback), and AI model evaluation services. Used by OpenAI, Google, Microsoft, and other leading AI labs, Scale powers the training of the world's most capable AI models.

AI platform for computer vision development

Roboflow is a platform that makes it easy to build, train, and deploy computer vision models. It handles dataset management, annotation, model training, and deployment in a unified workflow. Roboflow supports object detection, image classification, and instance segmentation for applications like quality control, security, and autonomous systems.

MLOps platform for AI model training and monitoring

Weights & Biases (W&B) is the leading MLOps platform for AI researchers and ML engineers. It tracks experiments, visualizes model performance, manages datasets, and monitors deployed models. W&B integrates with every major ML framework and is used by leading AI labs including OpenAI, NVIDIA, and Samsung.

Managed vector database for AI applications

Pinecone is the leading managed vector database service for AI applications. It stores and searches high-dimensional vector embeddings at scale, making it the backbone of RAG (Retrieval Augmented Generation) systems, semantic search, and recommendation engines. Pinecone is serverless and scales to billions of vectors.

Open-source AI-native vector database

Weaviate is an open-source, AI-native vector database that stores objects and their vector representations. It enables hybrid search combining vector similarity with traditional keyword search, making it ideal for complex RAG and AI application architectures. Weaviate supports automatic vectorization through module integrations.

Ultra-fast AI inference on wafer-scale chips

Cerebras is an AI hardware and inference company with its Cerebras Inference service offering some of the fastest LLM inference speeds available. Running Llama models at over 2,000 tokens per second, Cerebras enables real-time AI applications that require extremely fast response times and high throughput.

Cloud platform for AI application development

Lepton AI is a cloud platform for building and deploying AI applications at scale. It simplifies running AI models by handling infrastructure, auto-scaling, and GPU management. With a Pythonic SDK, developers can deploy AI workloads in minutes without DevOps expertise.

Run large language models locally on your computer

Ollama is the most popular tool for running open-source large language models locally on Mac, Linux, and Windows. It provides a simple CLI and API for downloading and running models like Llama 3, Mistral, Gemma, Phi, and hundreds more. Ollama makes local AI accessible without complex setup, enabling fully private, offline AI.

Desktop app to discover and run local LLMs

LM Studio is a user-friendly desktop application for discovering, downloading, and running open-source language models locally. It features a ChatGPT-like interface for local models, an OpenAI-compatible local server, and GPU acceleration on Apple Silicon and NVIDIA GPUs. LM Studio makes local AI accessible to non-developers.

Open-source local AI chatbot for any hardware

GPT4All is an open-source ecosystem for running powerful language models locally on consumer-grade hardware. It includes a desktop application, a suite of locally running models, and an open-source Python library. GPT4All allows document querying, local AI assistants, and private AI chat without internet.

Full-stack private AI workspace for documents and chat

AnythingLLM is an open-source, full-stack application for building private AI workspaces on any LLM. It supports local models via Ollama and cloud APIs, with built-in RAG for document Q&A, multi-user workspaces, and an agent system. AnythingLLM is the most feature-complete self-hosted AI workspace.

Open-source offline-first AI chat desktop app

Jan is an open-source, offline-first alternative to ChatGPT that runs 100% on your computer. It supports all major open-source models and connects to remote APIs like OpenAI and Anthropic. Jan focuses on privacy, extensibility, and a clean ChatGPT-like interface for local and remote AI.

Private, offline document Q&A with AI

PrivateGPT is an open-source AI project that allows querying documents using AI with full privacy — no data ever leaves your machine. Built on LlamaIndex and local LLMs, it ingests documents and answers questions about them completely offline. It is widely deployed in air-gapped enterprise environments.

Free, open-source OpenAI API replacement for local inference

LocalAI is a free, open-source alternative to the OpenAI API that runs locally. It provides an OpenAI-compatible REST API for text generation, image generation (Stable Diffusion), audio transcription (Whisper), and TTS — all running on consumer hardware without a GPU requirement. It is a drop-in replacement for OpenAI API calls.

Meta's most capable open-source language model family

Llama 3 is Meta's flagship open-source large language model family, available in 8B and 70B parameter sizes (and Llama 3.1 up to 405B). It matches or exceeds closed-source models on many benchmarks and can be run locally, fine-tuned, or deployed via commercial cloud providers. Llama 3 is the most widely deployed open-source LLM.

Chinese open-source frontier AI models

DeepSeek is a Chinese AI company producing frontier open-source language and code models. DeepSeek-V3 and DeepSeek-R1 match GPT-4 level performance at a fraction of the training cost, disrupting the AI industry. DeepSeek-R1 is an open-source reasoning model competitive with OpenAI's o1, making it the most significant open-source release in recent AI history.

Google's open-source lightweight language models

Gemma is a family of lightweight, open-source language models from Google built on the same research as Gemini. Available in 2B and 7B sizes, Gemma models are optimized for on-device and edge deployment. Gemma 2 improved significantly on reasoning, and PaliGemma adds vision capabilities.

Microsoft's small but powerful open language models

Phi-3 is Microsoft's family of small language models (SLMs) that punch far above their weight class. Phi-3-mini (3.8B) outperforms models 10x its size on reasoning benchmarks. The Phi series proves that carefully curated training data produces highly capable small models suitable for on-device and edge deployment.

Alibaba's open-source multilingual AI model family

Qwen (Tongyi Qianwen) is Alibaba Cloud's family of open-source language models ranging from 0.5B to 72B parameters. Qwen2.5 models are highly competitive with frontier models across coding, math, and multilingual tasks. The Qwen family includes specialized code, math, and vision-language models.

Cohere's enterprise-optimized RAG language model

Command R+ is Cohere's flagship language model optimized for retrieval augmented generation (RAG) and enterprise use. It excels at multi-step tool use, retrieval tasks, and grounded generation with citations. Command R+ supports 10 languages and is designed to be deployed in private cloud or on-premise environments.

Open-source LLM from the Technology Innovation Institute

Falcon is a family of open-source causal decoder-only language models from the Technology Innovation Institute (TII) in Abu Dhabi. Falcon 180B was one of the largest open-source models when released and remains a strong performer on academic benchmarks. Falcon models use a custom training setup with curated RefinedWeb data.

Multilingual open-source LLM by BigScience

BLOOM is a 176B parameter open-source multilingual language model created by the BigScience collaboration involving over 1,000 researchers. It was the first open-access LLM to match GPT-3 in scale and supports 46 human languages and 13 programming languages — a significant achievement for open multilingual AI.