Quick answer: OSS-model hosting + fine-tuning — cheap + fast inference for open models.
Together AI is an open-source model hosting and fine-tuning platform that provides cost-effective inference for open-weight large language models. Built by Together Computer, it offers a unified API for accessing models like Llama, Mistral, and other OSS alternatives to closed-source LLMs. The platform specializes in fast, affordable token-based pricing, making it ideal for developers and enterprises seeking to run inference at scale without the premium costs of proprietary models. Together AI handles the infrastructure complexity—GPU allocation, model optimization, batching—so you focus on building applications. It supports both immediate inference through their API and fine-tuning capabilities for customizing models on your own data.