Together AI

Together AI cloud GPU + inference open-source.

Définition

Together AI services : Serverless Inference (Llama 3.3 70B, Mistral, DeepSeek-V3, Qwen 2.5, Flux 1 image gen), Dedicated Endpoints, Fine-tuning (LoRA, full fine-tuning), Together GPU Clusters (H100, H200, B200, GB200), Together Code Interpreter, Together Code Sandbox. Forte adoption RAG + agents. Levée Series B 305M$ 2025 valorisation 3.3Mds$.

Origine

Together Computer Inc., fondée 2022 à San Francisco, Series A 102M$ 2023, Series B 305M$ 2025.

Usage

Concurrent Hugging Face Inference, Replicate, Fireworks AI, Anyscale. Force chez large-scale training cluster + inference combination.

Termes liés

• HF Inference Endpoints — concurrent.
• Replicate — concurrent serverless.
• Open-source LLM — catégorie modèles.