Databricks provides a unified Data + AI platform (lakehouse) for data engineering, analytics, and machine learning to build, train, and deploy models on cloud infrastructure.

Amazon SageMaker is AWS’s fully managed machine learning service for building, training, tuning, and deploying ML models at scale across the ML lifecycle.

Vertex AI (Google Cloud) is a managed, unified ML platform for training, tuning, deploying, and monitoring ML models and for MLOps workflows across the model lifecycle.

Azure Machine Learning is Microsoft’s enterprise-grade service for the end-to-end ML lifecycle: data prep, model training, MLOps, deployment, and governance across cloud and hybrid environments.

Hugging Face is a developer and model hub for open-source machine learning models, datasets, libraries (Transformers, Datasets, Tokenizers) and hosted inference/finetuning services.

Gradient (by Paperspace) is a cloud ML platform providing managed notebooks, distributed training, orchestration (Workflows), and deployment tools to run ML workloads on GPU/TPU infrastructure.

Google Colab is a hosted Jupyter notebook environment that requires no setup and provides free (and paid Pro/Pro+) access to CPUs, GPUs, and TPUs for machine learning, data analysis, and education.

Kaggle is a data science community platform offering competitions, datasets, notebooks, and learning resources for data scientists and ML practitioners.

Google Cloud AI Platform (Training & Prediction) is Google’s managed service for training and serving ML models; many capabilities have been consolidated under Vertex AI but AI Platform documentation and APIs remain for specific workloads.

IBM Watson Studio (now part of the watsonx.ai Studio experience) is IBM’s data science and AI platform that provides collaborative tools, notebooks, AutoAI, model deployment, and governance across cloud and on-prem environments.

Oracle’s AI & ML portfolio on Oracle Cloud Infrastructure (OCI) includes OCI Data Science, Generative AI services, ML services, and infrastructure designed for model training, tuning, deployment, and data-centric ML operations.

Dataiku is an end-to-end enterprise AI and data platform (Dataiku DSS) for data preparation, feature engineering, model training, deployment, and operationalization with collaboration and governance features.

RapidMiner is a data science platform (Studio, AI Hub/Server) providing a visual workflow designer, automated model building, collaboration/AI Hub, and deployment capabilities for predictive analytics and ML.

KNIME is an open-source analytics and data science platform (KNIME Analytics Platform) with visual workflow building, extensibility via nodes, and enterprise server capabilities for automation and deployment.

Domino Data Lab (Domino) is an enterprise MLOps/data science platform that centralizes environments, code, data, and model operations to accelerate research, reproducibility, deployment, and governance.

A developer-first MLOps platform for experiment tracking, dataset & model versioning, visualization, and collaboration across ML projects.

An open-source platform for the ML lifecycle that provides experiment tracking, a model registry, and deployment tools to productionize models.

An MLOps platform offering experiment management, model registry, production monitoring, and LLM/agent tracing and evaluation tools.

An experiment tracking and metadata platform focused on large-scale model training and deep debugging of model internals.

A cloud-agnostic MLOps platform that automates reproducible pipelines, run/version tracking, and orchestration across hybrid and multi-cloud infrastructure.

An open-source Kubernetes-native platform that provides composable tools for ML workloads including notebooks, pipelines, training operators, hyperparameter tuning, and model serving.

An open-source tool for data and model versioning, reproducible pipelines, and experiment management that integrates with Git and external storage backends.

An open-source full-stack platform for experiment tracking, orchestration, dataset management, and model serving with an enterprise control plane for infrastructure management.

An open-source toolkit for tracking experiments, automating runs, hyperparameter tuning, and pipeline automation aimed at simplifying ML experiment management.

A platform for automating, tracking, and reproducing deep learning and ML workflows that deploys on Kubernetes and supports experiment management, pipelines, and optimization.

A data-centric platform that provides Git-like versioning for data, immutable lineage, and data-driven pipelines built on Kubernetes for reproducible data processing and ML.

A modular MLOps and LLMOps framework (Seldon Core) and enterprise products for deploying, monitoring, and managing real-time ML/LLM inference at scale.

A unified inference platform and open-source framework for packaging, deploying, and scaling model inference APIs and multi-model pipelines on any cloud or Kubernetes.

An open-source platform for deploying, scaling, and managing machine learning APIs and inference workloads on cloud infrastructure.

A cloud-native PaaS for building, deploying, and governing ML and LLM applications with an AI gateway, observability, and enterprise controls for hybrid and on-prem deployments.

A cloud-native managed vector database and similarity search service for production AI applications, built to store and query embeddings at scale with low latency.

An open-source, AI-native vector database that provides vector and hybrid search, model/vectorizer modules, and GraphQL/REST APIs for semantic search and RAG workflows.

An open-source, cloud-native vector database and similarity search engine focused on high-performance ANN search, storage efficiency, and easy deployment (Docker/Kubernetes).

An open-source, cloud-native distributed vector database designed for high-performance ANN search and large-scale vector workloads, available as OSS and via Zilliz Cloud.

An open-source embedding database (ChromaDB) tailored for LLM applications, offering vector storage, retrieval, and developer-friendly client APIs for Python and JavaScript.

A research-grade open-source library (from Meta/Fair) for efficient similarity search and clustering of dense vectors, with CPU and GPU implementations of many ANN algorithms.

An open-source platform for large-scale search and AI applications that combines vector search, text search, and distributed ML inference to build real-time, relevance-driven applications.

An open-source PostgreSQL extension that adds a vector column type and similarity search operators to Postgres, enabling embedding storage and ANN/ENN queries within the database.

A vector search platform focused on e-commerce and product discovery that offers multimodal semantic search, merchandising controls, and managed cloud options.

A multimodal lakehouse and vector database built on the Lance columnar format that stores vectors alongside multimodal data and supports vector search, SQL, and training pipelines.

A cloud-native, Kubernetes-oriented open-source vector search engine designed for highly scalable and distributed approximate nearest neighbor (ANN) search.

Elasticsearch provides native vector field types and vector search capabilities within its distributed search and analytics engine, enabling hybrid vector + keyword/analytics queries.

Redis (with RediSearch/Redis Vector capabilities) offers in-memory vector similarity search and hybrid query capabilities for ultra-low-latency, high-throughput semantic and exact-match workloads.

Atlas Vector Search is MongoDB Atlas’s native vector search capability that stores and queries embeddings alongside operational data, enabling hybrid queries combining vectors, filters, and aggregation pipelines.

SingleStore is a modern distributed SQL database that provides native vector data types and indexed ANN vector search alongside high-performance SQL analytics and hybrid search capabilities.

A platform to run, share, and deploy machine learning models via a simple API and model repository; developers can run community and custom models with one line of code.

A cloud platform for running ML workloads (inference, training, batch jobs) with fast cold-starts, autoscaling, and a developer-oriented programmable infrastructure.

Not verified

Apache Beam is an open-source unified model for batch and streaming data processing; its RunInference API enables integrating ML model inference into Beam pipelines for large-scale, streaming or batch inference.

A cloud platform built for AI that provides GPU infrastructure to train, deploy, and scale models with serverless and managed GPU options.

Together AI is an AI acceleration cloud offering GPU clusters, model inference (serverless or dedicated), and fine-tuning tools for training and running frontier models.

Anyscale is a managed platform built around Ray that enables developers to run, scale, and operate distributed ML and AI workloads (data processing, training, inference) from laptop to cloud.

Ray Serve is a scalable, framework-agnostic model serving library built on Ray for building online inference APIs, with features tailored to large-model serving and model composition.

An open-source model serving framework for PyTorch models that provides REST/gRPC inference endpoints, model management, and production-oriented features (note: project documentation indicates limited maintenance status).

TensorFlow Serving is a flexible, high-performance serving system designed for production that provides versioned model serving, gRPC/HTTP endpoints, and batching for TensorFlow (and extendable to other model types).

A modern, high-performance Python web framework for building APIs, with automatic OpenAPI documentation, data validation using Pydantic, and async support (commonly used to expose model inference endpoints).

A Python library to quickly build and share interactive web UIs for machine learning models (demos and simple apps), with easy hosting and embedding workflows.

An open-source Python framework to build and share interactive data apps and model demos quickly using pure Python scripts, with hosting options via Streamlit Community Cloud.

A hosting platform on Hugging Face for ML demo apps and interactive model interfaces (supports Gradio, Streamlit, Docker, or static apps) and offers GPU/runtime upgrades for ML workloads.

A platform focused on production inference: deploy, optimize, and scale open-source, custom, and fine-tuned AI models with an inference-optimized stack and dev experience for mission-critical deployments.

A developer platform and API for OpenAI’s flagship LLMs and related capabilities (Responses, Assistants, Realtime, fine-tuning, tools, and model extensions) to build conversational agents, embeddings, and multimodal apps.

Anthropic’s developer console for accessing the Claude family of models and related developer tooling (API keys, docs, model list and management) to build and deploy Claude-based applications.

Google’s web-based IDE and developer platform for prototyping and building with Google’s Gemini/multimodal models, offering prompt tools, a prompt gallery, API key management and a path to Vertex AI for production.

A managed AWS service that provides API access to multiple foundation models (including partner and AWS models) plus developer features for RAG, fine-tuning, guardrails, agents, and integration with SageMaker Unified Studio.

Microsoft Azure’s managed OpenAI service that provides access to OpenAI models (Responses/Chat/GPT family) and Azure-specific tooling (studio, enterprise security, private networking, and corporate compliance integrations).

Cohere provides LLM APIs and an enterprise AI platform with generation, embeddings, reranking, and model customization (including private deployments and VPC options) focused on secure, scalable business use.

AI21 Labs’ developer platform (AI21 Studio) for accessing Jurassic/Jamba family models and task-specific APIs, plus tools like file libraries, SDKs, and deployment options (SaaS, cloud marketplace, or self-hosted).

Together AI provides a GPU cloud platform and managed inference/fine-tuning services for open-source and frontier models, plus instant GPU clusters, optimized inference engines, and tooling for training and deployment.

Fireworks offers a managed inference cloud optimized for fast, low-latency inference of open-source generative models and full model lifecycle management (run, tune, scale) with enterprise security and global distribution.

Anyscale Endpoints is Anyscale’s managed LLM serving offering that provides scalable, production-grade API endpoints for open-source LLMs with features like JSON mode, function calling, private endpoints, and integration into the Anyscale platform.

Replicate provides a cloud API to run machine learning models (image, video, audio, LLMs) without managing infrastructure — enabling running community-published models, deploying custom models, and fine-tuning where supported.

Hugging Face Inference (Inference API and Inference Endpoints) provides hosted inference for models from the Hugging Face Hub, plus dedicated endpoints, Inference Providers integrations, and SDKs to run language, vision, and multimodal models.

Groq develops inference-optimized AI accelerators (LPUs) and provides hardware-backed inference solutions and cloud offerings that accelerate LLM inference workloads for high-throughput, low-latency production usage.

Perplexity’s developer APIs (pplx-api / Sonar) provide fast inference access to open-source LLMs and Perplexity’s grounding/search features, enabling programmatic AI search, retrieval-augmented answers, and hosted open-model inference.

Mistral AI’s API provides programmatic access to Mistral’s family of models (text, code, vision) with endpoints for chat, completions (including fill-in-the-middle), embeddings, fine-tuning, function calling, and structured outputs.

OpenAI’s fine-tuning API lets developers customize OpenAI models (including GPT‑4o and GPT‑4o mini) on their own datasets for text and image+text tasks, and supports workflows like model distillation and a fine-tuning dashboard. Documentation and product announcements describe API endpoints, data-format guidance, and privacy/ownership guarantees for fine-tuned models.

Lamini is an enterprise LLM platform that provides tooling and an engine to train, specialize, and deploy LLMs on proprietary data, with options for on‑premises or VPC deployment and support for multi‑node training (including AMD GPU optimizations).

Predibase is a platform for fine-tuning, reinforcement fine-tuning, and serving open-source LLMs; it provides a Python SDK, UI, serverless production endpoints and optimizations (LoRAX) to enable memory-efficient training and cost-effective serving of fine-tuned models.

Anyscale provides a managed platform and optimizations around Ray Train (the Ray library for distributed training) — enabling distributed/fault-tolerant model training, elastic training, autoscaling, checkpointing and integrations with common ML frameworks (PyTorch, TensorFlow, Hugging Face).

Together.ai offers a GPU cloud and platform for training, fine-tuning (full and LoRA) and serving models with ready clusters, an optimized software stack (custom kernels like FlashAttention-3), and APIs/CLIs to submit and manage fine-tuning jobs and deployments.

Weights & Biases (W&B) is an experiment tracking and MLOps platform that logs training runs, artifacts, datasets and metrics; it provides model & dataset registries, monitoring, and integrations for fine-tuning and distributed training workflows.

Lit-GPT is Lightning AI’s open-source repository and toolset (GitHub + docs) for implementing, pretraining and fine‑tuning transformer LLM architectures; it includes scripts and support for LoRA/QLoRA, FlashAttention, quantization, and conversion utilities for popular open-source checkpoints.

Axolotl (OpenAccess-AI-Collective) is an open-source fine‑tuning framework that streamlines fine‑tuning of many open-source LLM families, offering YAML-based configs, LoRA/QLoRA/Relora/full-finetune support, multi‑GPU deployment (FSDP/DeepSpeed), and integrations with experiment loggers.

LM Studio is a cross‑platform desktop application and model catalog for running and managing local/open-source LLMs (inference, model catalog with model.yaml, local RAG and plugins). Official docs describe local/offline model execution, model.yaml model metadata, and local hosting; LM Studio’s documentation centers on inference and model management rather than on fine‑tuning workflows.

GPT4All (Nomic and community repos/docs) is an open-source ecosystem and local runtime that provides desktop apps, CLI and Python bindings for running quantized LLMs locally, plus tooling to download and run community models; documentation covers local generation APIs and desktop installers.

A marketplace and library for buying, selling, and discovering tested AI prompts (templates) across models and use cases. ([promptbase.com](https://promptbase.com/?utm_source=openai))

An AI-powered prompt optimizer and prompt-as-a-service product by Jina AI that rewrites and optimizes user prompts for different LLMs and image models and exposes APIs/SDKs for integration. ([pypi.org](https://pypi.org/project/jinaai/?utm_source=openai))

A prompt engineering IDE that lets teams compose prompts from modular blocks, test prompt reliability, run evaluators, and collaborate with versioning and spend tracking. ([promptmetheus.com](https://promptmetheus.com/?utm_source=openai))

A developer-first tool for prompt testing, red teaming, and automated evaluations (CLI and managed) that focuses on finding prompt-level security, reliability, and safety issues. ([promptfoo.dev](https://www.promptfoo.dev/?utm_source=openai))

LangChain Hub (accessible via LangSmith) is a repository and discovery hub for prompts (and other artifacts like chains/agents) where users can browse, download, edit, and commit prompt templates to LangChain workflows. ([blog.langchain.com](https://blog.langchain.com/langchain-prompt-hub?utm_source=openai))

A prompt management and LLM observability platform that provides a prompt registry, visual editor, versioning, A/B testing, and evaluation pipelines to iterate prompts with non-technical stakeholders. ([promptlayer.com](https://www.promptlayer.com/?utm_source=openai))

An open-source LLM observability and gateway platform that logs requests, provides analytics, caching/routing, prompt experiments, and a unified API to monitor model performance and cost. ([helicone.ai](https://www.helicone.ai/?utm_source=openai))

Humanloop was a development platform for building, managing, and evaluating LLM applications; the company announced that its team joined Anthropic and that the Humanloop platform is being sunset, with migration guidance provided. ([humanloop.com](https://humanloop.com/))

An evals, observability, and AI-engineering platform focused on building reliable AI agents: Braintrust provides playgrounds, evals, automated scoring, production monitoring, and tools to iterate and ship prompts and agent workflows. ([braintrust.dev](https://www.braintrust.dev/?utm_source=openai))

A full-stack LLMOps platform (AI Gateway, observability, guardrails, governance, and prompt management) for creating, testing, and deploying production-ready prompts and workflows across many models. ([portkey.ai](https://portkey.ai/?utm_source=openai))

OpenAI Evals is an open-source framework and registry for creating, running, and sharing evaluations of large language models (LLMs) and LLM systems. It provides templates, built-in benchmarks, and tooling to run, score, and compare model outputs or build custom evals for specific use cases.

Weights & Biases is an ML platform for experiment tracking, model and dataset versioning, evaluation, and observability across the model lifecycle. It includes tools for logging experiments, comparing model versions, tracing model inputs/outputs, and purpose-built evaluation tooling (Weave) for LLMs and agentic systems.

Arize AI is an ML observability platform that monitors model performance in production, detects data and model drift, and provides analytics and explainability to diagnose and remediate issues. It connects training, validation and production data to help teams investigate regressions and optimize models over time.

WhyLabs provides an AI and data observability platform (with open-source components like whylogs) that captures statistical profiles of datasets and model I/O to detect anomalies, drift, and data-quality issues while preserving privacy. The platform includes an AI Control Center for observing, securing, and optimizing ML and generative AI applications.

Fiddler AI is an AI observability and explainability platform that provides continuous monitoring, root-cause analysis, and explainable predictions for both classical ML and LLM applications. The platform includes guardrails, fairness checks, explainers (e.g., SHAP/gradients), and reporting features for governance and audit evidence.

Arthur AI is a model monitoring platform that proactively monitors model performance, detects bias and anomalies, and provides explainability for tabular, NLP, and computer vision models. It offers dashboards, alerts, and tools to debug model regressions and operationalize model health checks.

Evidently AI is an open-source library and platform for evaluating, testing, and monitoring machine learning and LLM systems. It offers 100+ built-in metrics, drift detection, interactive reports and a platform for no-code/CI-driven evaluations and production monitoring.

Deepchecks offers a library and platform for automated testing and validation of ML models and data, including suites for data integrity, distribution checks, model performance, and LLM/agent evaluation. It supports pre-deployment tests, CI pipelines, and production monitoring for traditional ML and LLM-based applications.

Great Expectations is an open-source data quality and validation framework (with a commercial cloud offering) that lets teams define 'Expectations' — verifiable assertions about data — and run validations, checkpoints and generate human-readable Data Docs. It is used to test, document and monitor data quality and schemas across pipelines.

Giskard is an open-source ML testing framework and commercial platform for detecting vulnerabilities and testing AI models (from tabular models to LLMs). It provides automated scans, test-generation, a collaborative hub for domain experts, and tools for red-teaming LLMs and building test suites to prevent regressions.