Inference.ai

Inference.ai is a platform that groups four products around running AI workloads: Ghost, Maestro, Engine and Academy. It is aimed at developers, startups and enterprises that want to deploy agents, route requests across models, rent GPUs, or learn AI by building. Ghost deploys an agent on a persistent virtual machine with one click. The vendor says everything is configured automatically and the machine is warm in about 60 seconds. Maestro is a routing layer that picks a model for each request based on cost, speed and quality, with hard budget caps and a single unified bill. The site marks Maestro as coming soon. Engine sells wholesale bare metal GPU compute, from Blackwell B300 down to RTX 4090 cards, billed hourly or reserved, with InfiniBand clusters that scale to 4,096 GPUs. Academy teaches AI by having learners build on the same infrastructure, with a coaching assistant called TAi that follows the learner's reasoning and asks follow up questions. The platform also hosts open models for users, including families from DeepSeek, Qwen, Kimi, GLM and MiniMax, and describes access to frontier models through one API key. The site does not publish plan prices. The only dollar figures on the page are sample per request costs inside the Maestro routing illustration. Sales contact is offered for Engine and enterprise needs, and a Deploy an Agent button starts the Ghost flow.