Skip to content

Companies

The methodology employed categorizes companies based on their primary focus or offerings, detailing each company's capabilities alongside its name.

Key to Keywords:
- Pre: Pretraining capabilities or services
- FT: Finetuning services
- Custom: Support for building custom models
- API: Offers an API to integrate ML functionality
- Dev: Development tools or environments
- Dep: Deployment tools or platforms
- Model: Develops or provides their own ML models
- GPU: GPU-based compute resource provider
- Storage: Data storage provider
- Data: Data platform or data indexing provider

Highlights

Models are accessed via their respective companies, such as OpenAI (ChatGPT 40, o1, o3), DeepSeek (DeepSeek-R1), Anthropic (Claude Sonnet), and Google (Gemini). LM Studio is a desktop application for running models locally. Together AI provides a platform for running models that might not fit on your local machine. There are other serving locations, such as the Hugging Face Inference Playground, and Hyperbolic.xyz, which also offers base models.

Model Creators / Foundation Model Providers

API/Inference Providers

Development Tools, Code Generation, and Productivity

Platforms like Bolt, Replit Agent, Vercel V0 use agentic workflows to improve code quality. They also help deploy generated applications.

Other tools related:

  • Llama Stack developed by Meta is a collection of APIs that standardize the building blocks necessary for developing generative AI applications.
  • Adept's ACT-1: Adept AI has developed ACT-1, an AI model designed to interact with software applications through natural language commands. It can perform tasks such as navigating web pages, clicking buttons, and entering data, effectively acting as a digital assistant to automate workflows.

Deployment Tools and Platforms

  • Vercel: Dev, Dep (V0 by Vercel generates UI from text)
  • Open WebUI: Dep (Manage and deploy AI models locally. UI for interacting with various LLMs. Integrate with LLM runners like Ollama and OpenAI-compatible APIs)

Libraries and Frameworks

  • Ollama, vLLM: tool designed to run LLMs locally. It can be used with Chatbox.ai app desktop, mobile and web-based app.
  • unsloth (Daniel Han), torchtune, Oxolotl: Fast Training (Pre/FT related). They enhance the speed and efficiency of LLM fine-tuning.
  • LiteLLM: LiteLLM is an open-source Python library designed to streamline interactions with a wide range of LLMs by providing a unified interface.
  • ai-gradio: A Python package that makes it easy for developers to create machine learning apps powered by various AI providers. Built on top of Gradio, it provides a unified interface for multiple AI models and services.
  • DSPy: Dev (Provides a programming model for developing and optimizing language model pipelines)
  • Torchrun: Dev
  • Cog (by Replicate): Packaging Custom ML Models for Deployment (Custom, Dep)
  • ComfyUI: Dev (GUI for Workflow)
  • Tinygrad: Dev
  • Agentic AI systems: LangChain, CrewAI, LlamaIndex, Haystack, Devin
  • OpenHands: Dev (providing frameworks for building AI-driven applications with agentic capabilities)

Data Platforms and Providers

End-to-End Cloud-Based ML Platforms