Crusoe Gives Organizations More Granular Control Over Open AI Models
Techstrong.ai, Tuesday, July 7th, 2026
Crusoe adds serverless fine-tuning and self-serve deployments to give organizations more control over open models.
Mike Vizard reports Crusoe launched two capabilities in its Intelligence Foundry platform. Serverless fine-tuning lets organizations pick open-weight base models, upload datasets, and create optimized inference models, with automatic recovery on hardware failure and portable safetensors weights deployable anywhere.
Self-serve deployments let teams choose throughput- or responsiveness-optimized inference profiles on NVIDIA H100 or H200 GPUs, billed hourly.
SVP Erwan Menard says organizations increasingly prioritize control and cost over model origin. Crusoe supports Qwen, DeepSeek, Gemma, and gpt-oss; the neocloud sector is about 10% of the market amid GPU scarcity.