Replicate
ReplicatePlatform for running thousands of open and proprietary AI models through an API, and for deploying your own with Cog.
Run open language models on your own computer with one command, or call larger open models in the cloud through the same interface.
Ollama is a model runner for open-weight language models. After installing it on macOS, Windows or Linux, one command downloads a model such as Gemma, gpt-oss, Mistral or DeepSeek and starts a local chat or a local API that other apps can call. Everything that runs on your own hardware is free and unlimited, which makes it a common choice for developers who want private, offline AI or who build tools on top of a local model. A newer cloud option runs larger open models on Ollama's servers with per-token pricing and the same commands, so a project can move between a laptop and the cloud without code changes. It is aimed at developers, researchers and privacy-minded users rather than people who want a polished chat app; the difference from hosted chatbots is that you choose the model and where it runs.
Platform for running thousands of open and proprietary AI models through an API, and for deploying your own with Cog.
Generative media platform with fast APIs for hundreds of image, video and audio models plus serverless GPUs for custom ones.
Google's free web workspace for trying Gemini models, building prompts and apps, and getting a Gemini API key.
Code completion and agents inside the editors most teams already use