Fornax · Models & APIs
AI models & APIs
AI models & APIs in this graph: 66, including TensorFlow, Transformers, PyTorch, CUDA, Hugging Face and MarkItDown.
Where builders get models: APIs, open weights and local runtimes.
Capabilities in this galaxy
- LLM API23
- Model training16
- Fine-tuning15
- Model serving15
- Run models locally12
- Vector search10
- Knowledge base (RAG)9
- Document parsing8
- Find and share models7
- One API for many models7
- Text recognition (OCR)7
- Embeddings6
- Cloud GPUs5
- Speech to text4
- Ask your documents3
- Build AI agents3
- Chat and Q&A3
- Evaluate and trace LLM apps3
- Build a chatbot2
- Deploy and host2
- Prompt optimization2
- Text to image2
Products15
- Hugging Face2016The home of open models, datasets and demos.Free plan + paid upgrades · Runs offline
- Ollama2023Run open models on your own machine with one command.Open source, free · Runs offline
- ModelScope2022Alibaba's model hub and library for finding, running and fine-tuning open models.Open source, free · Runs offline
- Claude Developer Platform2023APIs for Claude models, tool use and agent building.Paid
- Google AI Studio2023Try Gemini models for free and get an API key in minutes.Free
- OpenAI Platform2020APIs for OpenAI's models, agents and realtime voice.Paid
- OpenRouter2023One API for hundreds of models, routed by price and uptime.Free plan + paid upgrades
- One API2023A self-hosted gateway that puts many LLM APIs behind one OpenAI-compatible endpoint, with key management.Open source, free
- Langfuse2023An open-source platform for tracing, evaluating and monitoring LLM apps and agents.Open source, free
- fal2021Fast inference for image, video and audio models.Paid
- Groq2024Extremely fast inference on custom LPU chips.Free plan + paid upgrades
- LM Studio2023A desktop app to find, download and chat with local models.Free · Runs offline
- Replicate2021Run and fine-tune open models through a simple API.Paid
- Together AI2022A cloud for running, fine-tuning and training open models.Paid
- Modal2023Serverless GPUs for running models and batch jobs from Python.Paid
CLIs1
Frameworks49
- TensorFlow2015Google's open-source machine learning framework for building and deploying models across devices.Open source, free · Runs offline
- Transformers2018Hugging Face's library of pretrained model definitions for text, vision and audio, used for inference and training.Open source, free · Runs offline
- PyTorch2016An open-source deep learning framework with dynamic computation graphs, first developed at Meta AI.Open source, free · Runs offline
- CUDA2007NVIDIA's parallel computing platform and programming model for general-purpose computing on its GPUs.Runs offline
- MarkItDown2024A Python tool that converts PDFs, Office files, images and more into Markdown for LLMs.Open source, free · Runs offline
- llama.cpp2023A C/C++ engine by Georgi Gerganov for running quantized LLMs locally on CPUs, Apple silicon and GPUs.Open source, free · Runs offline
- vLLM2023A high-throughput engine for serving LLMs, built around the PagedAttention memory technique.Open source, free · Runs offline
- PaddleOCR2020An OCR and document parsing toolkit that turns images and PDFs into text, tables and structure.Open source, free · Runs offline
- MinerU2024A tool that parses PDFs, images and Office files into Markdown and JSON ready for LLMs.Open source, free · Runs offline
- JAX2018A Google library for composable transformations of NumPy programs: autodiff, JIT compilation and vectorization.Open source, free · Runs offline
- Unsloth2023A framework and local app for fine-tuning and running LLMs faster and with less memory.Open source, free · Runs offline
- LlamaFactory2023A toolkit with a web UI for fine-tuning over a hundred language and vision models.Open source, free · Runs offline
- Docling2024A library that parses PDFs, Office files and other documents into structured data for AI pipelines.Open source, free · Runs offline
- nanoGPT2022A minimal PyTorch codebase for training and fine-tuning GPT-2-sized models.Open source, free · Runs offline
- LiteLLM2023A proxy server and Python SDK that call over 100 LLM APIs in one OpenAI-compatible format.Open source, free
- PrivateGPT2023An API layer for building private AI apps, such as document Q&A, on local models.Open source, free · Runs offline
- LocalAI2023A self-hosted engine that serves LLMs, speech, image and video models behind an OpenAI-compatible API.Open source, free · Runs offline
- exo2024A runtime that joins several of your own devices into a cluster to run large models locally.Open source, free · Runs offline
- Milvus2019A cloud-native vector database for similarity search over billions of vectors.Open source, free · Runs offline
- Streamlit2019A Python framework that turns scripts into interactive web apps for data and AI.Open source, free · Runs offline
- Gradio2018A Python library for building and sharing web demos of machine learning models.Open source, free · Runs offline
- DeepSpeed2020A library for training and running very large models across many GPUs.Open source, free · Runs offline
- Faiss2017A C++ and Python library for fast similarity search and clustering of dense vectors.Open source, free · Runs offline
- Marker2023A tool that converts PDFs and other documents to Markdown, JSON or HTML.Open source, free · Runs offline
- LightRAG2024A lightweight RAG framework that retrieves through a knowledge graph as well as vectors.Open source, free · Runs offline
- SGLang2024An open-source serving framework for LLMs and vision-language models focused on fast structured generation.Open source, free · Runs offline
- GraphRAG2024A pipeline that builds a knowledge graph from your text and uses it for retrieval-augmented generation.Open source, free
- Qdrant2020A vector database and search engine for storing and querying embeddings at scale.Open source, free · Runs offline
- Chroma2022A database for storing and searching embeddings, with vector and full-text search.Open source, free · Runs offline
- MLX2023Apple's array framework for machine learning on Apple silicon, with a NumPy-like API.Open source, free · Runs offline
- AI SDK2023A TypeScript toolkit for building AI apps and agents with any model.Open source, free
- llamafile2023A way to ship an open model as a single executable file that runs locally without installing anything.Open source, free · Runs offline
- MLC LLM2023A compiler and engine that deploys LLMs natively on GPUs, phones and in the browser.Open source, free · Runs offline
- pgvector2021A Postgres extension that stores vectors and finds nearest neighbours.Open source, free · Runs offline
- PEFT2022A library for fine-tuning large models by training only a small set of extra parameters.Open source, free · Runs offline
- Surya2024An OCR toolkit for text detection, layout, reading order and tables in over 90 languages.Open source, free · Runs offline
- TRL2020A library for post-training language models with SFT, DPO, GRPO and other methods.Open source, free · Runs offline
- WebLLM2023A JavaScript engine that runs LLMs entirely in the browser with WebGPU.Open source, free · Runs offline
- Sentence Transformers2019A Python library for computing embeddings and training rerankers for search and retrieval.Open source, free · Runs offline
- Megatron-LM2019A library for training transformer models at scale across thousands of GPUs.Open source, free
- Weaviate2016A vector database that combines vector search with keyword filtering in one query.Open source, free · Runs offline
- Ragas2023A framework for evaluating LLM apps and RAG pipelines and generating test sets for them.Open source, free · Runs offline
- Unstructured2022A library that turns PDFs, images and Office files into clean, chunked data for LLMs.Open source, free
- TensorRT LLM2023A library for optimizing and serving LLM inference on NVIDIA GPUs.Open source, free
- Axolotl2023A framework for fine-tuning and post-training LLMs from a single YAML config.Open source, free · Runs offline
- Chainlit2023A Python framework for building chat interfaces for LLM apps.Open source, free
- Phoenix2022An open-source tool for tracing, evaluating and debugging LLM apps and agents.Open source, free · Runs offline
- LanceDB2023An embedded vector database for multimodal AI search, built on the Lance columnar format.Open source, free · Runs offline
- LMDeploy2023A toolkit for compressing, deploying and serving LLMs with fast inference engines.Open source, free
Models1
Organizations31
- Hugging Face2016The company behind the Hugging Face Hub for sharing open models and the Transformers library.
- NVIDIA1993The chipmaker whose GPUs and CUDA software supply most of the compute used to train AI.
- Amazon1994The retail and cloud company whose AWS arm offers Bedrock, the Nova models and Trainium chips.
- AMD1969A chipmaker that designs CPUs and the Instinct GPUs used for AI training and inference.
- Apple1976The maker of the iPhone and Mac, which designs its own chips and the Apple Intelligence models.
- Cloudflare2009A global network company offering CDN, security and a serverless developer platform, including hosting for AI agents.
- Huawei 华为1987A Chinese telecom and electronics company that makes Ascend AI chips and the Pangu models.
- Stripe2010A payments infrastructure company whose APIs let businesses accept payments and manage billing online.
- Vercel2015A frontend cloud platform company that makes Next.js, the AI SDK and v0.
- Chroma2022The company behind Chroma, an open-source embedding and vector database for AI applications.Open source, free
- Cohere2019A Toronto AI company that builds Command, Embed and Rerank models for enterprises.
- fal2021A generative media platform that serves image, video and audio models through an API.
- Firecrawl2022A company whose API crawls and scrapes websites into clean, LLM-ready Markdown and structured data.
- Groq2016A chip company that designs the LPU and runs a fast inference cloud on it.
- HashiCorp2012The company behind Terraform, Vault and other infrastructure-as-code tools, now part of IBM.
- MongoDB2007The company behind the MongoDB document database and the Atlas cloud database service.
- Ollama2023The company behind Ollama, a tool for running open models on your own computer.
- OpenRouter2023The company behind OpenRouter, a unified API for calling models from many providers.
- Replicate2019The company behind Replicate, a platform for running models through a cloud API.
- Scale AI2016A company that supplies labeled training data and evaluation services to AI labs.
- Supabase2020An open-source backend platform built on Postgres, with auth, storage and edge functions.Open source, free
- Together AI2022A cloud company for training, fine-tuning and serving open models.
- Apify2015A Prague-based web scraping and automation platform with a marketplace of ready-made scrapers called Actors.
- Element Labs2023Brooklyn company founded by Yagil Burowski that develops the LM Studio desktop app.
- Grafana Labs2014The company behind Grafana dashboards and open-source observability tools such as Loki and Tempo.
- Modal2021A serverless cloud for running AI and data workloads on GPUs from Python code.
- Neon2021A serverless Postgres company that separates storage from compute and supports database branching.
- Prefect2018A workflow orchestration company that maintains FastMCP, a Python framework for MCP servers.
- Pydantic2023The company behind the Pydantic data validation library, Pydantic AI and the Logfire observability platform.
- Qdrant2021A Berlin company that develops Qdrant, an open-source vector search engine written in Rust.Open source, free
- Upstash2020A serverless data platform offering Redis, vector and messaging services, and the maker of Context7.
People3
- Jensen Huang1993Co-founder of NVIDIA, whose GPUs became the main hardware for training AI models.
- Guillermo Rauch2015Founder of Vercel, the company behind Next.js and v0, and creator of Socket.IO.
- Simon Willison2022Developer and writer, co-creator of Django, known for documenting prompt injection and LLM tooling.
Hardware10
- NVIDIA H1002022NVIDIA's Hopper data-center GPU with a Transformer Engine, central to the generative AI build-out.
- Google TPU2016Google's custom tensor processing unit chips for training and serving neural networks.
- NVIDIA A1002020NVIDIA's Ampere data-center GPU, widely used to train large language models from 2020.
- NVIDIA Blackwell2024NVIDIA's GPU architecture after Hopper, used in the B200 and GB200 systems.
- AMD Instinct2016AMD's line of data-center GPU accelerators for AI and supercomputing, such as the MI300X.
- Apple silicon2020Apple's in-house Arm chips, such as the M series, whose unified memory suits running models locally.
- AWS Trainium2020Amazon's custom AI training chips, offered through AWS and used to train Claude.
- Groq LPU2020Groq's Language Processing Unit, a chip designed for low-latency model inference.
- Huawei Ascend 昇腾2018Huawei's family of AI processors for training and inference, China's main domestic alternative to NVIDIA GPUs.
- NVIDIA V1002017NVIDIA's Volta data-center GPU, the first with Tensor Cores for deep learning.
Last updated 2026-09-24