Frontier Model Companies
| Company | Model | Chat Site | Overview |
|---|---|---|---|
| OpenAI | ChatGPT | https://chatgpt.com | A leading AI company, notable for leading the current AI wave with the launch of ChatGPT in November 2022. GPT-4o is its top model; it is rumored to be training GPT-5. Many people consider this to be best LLM on the market. |
| Gemini | https://gemini.google.com | An earlier AI leader. Acquired DeepMind, which has made many AI breakthroughs. Google scientists authored the key paper Attention is all You Need. Makes its own AI chips (TPUs); owns a massive compute infrastructure and data. Gemini: large model family; Gemma: small model family. | |
| xAI | Grok | https://x.com/i/grok | Grok 2 beta iavailable in paid versions of X (Twitter). Just raised $6B in capital; access to large amount of data; led by a co-founder of OpenAI. |
| Meta | Llama | https://www.meta.ai | Makes Llama family of models, partially open source. Massive compute infrastructure and data. |
| Anthropic | Claude | https://claude.ai | AI research and products company let by former OpenAI employees. This year released Claude 3 - its top version was Opus, until they release Sonnet 3.5 (and upgraded it with the same name). Current leader in code generation with Artifacts. |
| Mistral | Mistral/Mixtral | https://chat.mistral.ai | Open generative AI. Mistral Large is a frontier model; also makes capable, smaller models. Codestral for generating code. |
Other Key AI Organizations
| Company | Site | Overview |
|---|---|---|
| Apple | https://www.apple.com/apple-intelligence | Apple Intelligence announced June 10, 2024. Tiered generative AI - smaller tasks on phone/computer, larger tasks in the cloud, where they are partnering with OpenAI. See this excellent overview: https://machinelearning.apple.com/research/introducing-apple-foundation-models. |
| Perplexity | https://www.perplexity.ai | Search service leveraging multiple LLMs. |
| Chatbot Arena | https://chatbotarena.com | Side by side answers, comparisons of features and pricing; includes both large and lesser known LLMs. |
| Nvidia | https://www.nvidia.com/en-us | Maker of the GPU chips that are fueling AI growth for most top AI companies. Became one of the top three companies by market cap earlier this year due to extremely high demand for its GPUs. Just released the Nemotron family of models. |
| Amazon | https://aws.amazon.com/ai | Top cloud infrastructure provider (Microsoft Azure is #2 and Google GCP is #3). Offers AWS Bedrock, a service for companies to build their own AI without the risk of exposing their data to the foundation companies. Also makes Atlas family of models. |
| Microsoft | https://www.microsoft.com/en-us/ai | Top backer of OpenAI; ChatGPT runs on Azure. Makes Phi small model family. Aggressively incorporating AI throughout their product offerings under the Copilot brand. |
| Hugging Face | https://huggingface.co | An AI development hub for sharing models, demos, and software libraries. |
|
Allen Institute for AI (AI2) Epoch AI |
https://allenai.org https://epochai.org |
AI research institutes. |
| National AI Advisory Committee (NAIAC) | https://ai.gov/naiac/ | The NAIAC is tasked with advising the President and the National AI Initiative Office on topics related to AI. |
Key AI Researchers/Commentators
| Who/What | Site | Overview |
|---|---|---|
| Interconnects (Substack) | https://www.interconnects.ai | “Long-form, technical AI analysis” |
| AI Snake Oil (Substack) | https://www.aisnakeoil.com | “What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference” |
| Francois Chollet | https://fchollet.substack.com/?utm_source=substack&utm_medium=web&utm_campaign=substack_profile | “I’m a software engineer and AI researcher. I started Keras, a deep learning framework, and ARC, a benchmark for general AI reasoning abilities. I’m also the author of Deep Learning with Python, a popular textbook to get started with deep learning.” |