Thunder Compute: Features second-billed persistent SSH GPU instances, but lacks a distributed consumer hardware backbone. Salad 记录了 713,519 月访问量,月度环比增长为 +10.4%,定价为免费试用,付费,定价请咨询。Thunder Compute 记录了 102,147 月访问量,增长率为 -29%,定价为 Paid (per-minute instances, per-second sandboxes)。

低成本云 GPU:面向 AI 工作负载的按分钟实例与按秒计费沙箱。
| 维度 / 指标 | ![]() | ![]() |
|---|---|---|
| 官方域名 | salad.com | thundercompute.com |
| 定价模式 | 免费试用,付费,定价请咨询 | Paid (per-minute instances, per-second sandboxes) |
| 月访问量 | 71.35万 (713,519) | 10.21万 (102,147) |
| 月度环比流量变化 | +6.75万 (+10.4%) | -4.17万 (-29%) |
| 主要市场与份额 | 🇺🇸 美国 (40.9%) | 🇺🇸 美国 (53.5%) |
| 直接访问 / 月 | 44.09万 | 4.07万 |
| 用户评分 | 4.3 | 4.4 |
| 支持的平台 | 网站 | 网站 |

编排全球闲置消费级 GPU 的分布式 GPU 云
在数千个分布式 GPU 节点上调度、监控并重新分配 Docker/OCI 工作负载,节点消失时自动故障转移。
按需访问覆盖 190 多个国家的闲置消费级 GPU 分布式算力池,优先定价约从每小时 0.015 美元起。
面向开源权重模型的托管 token 端点,按 token 计费,你的数据不会被用于训练。
在组织内部运行同一套编排栈,把闲置工作站 GPU 汇聚成内部兼容 OpenAI 的服务。

低成本云 GPU:面向 AI 工作负载的按分钟实例与按秒计费沙箱。
开通 RTX A6000、L40、A100 80GB 或 H100 SXM 机器,通常几分钟内即可 SSH 登录,只在运行时计费。
启动带或不带 GPU 的隔离 Ubuntu 虚拟机,用于 RL rollout、评估与 CI,按秒计费,任务之间可缩容至零。
CLI、VS Code 扩展、MCP 服务器与 REST API,让你能从终端、编辑器、AI Agent 或自有服务驱动基础设施。
将实例磁盘保存为快照并在之后恢复,实验可随时暂停,无需让机器持续运行。

专项更强专门提供开源模型的托管式无服务器推理 API,而非单纯租赁 GPU 基础设施。
从全球市场按秒租用云 GPU
免费 / 开源面向 AI 与数据负载的无服务器云基础设施
Nebius is an AI-focused cloud platform that gives teams non-virtualized NVIDIA GPU clusters, managed MLOps tooling, and a serverless inference layer so they can move from experiment to production on one stack.
Nebius is an AI-focused cloud platform that gives teams non-virtualized NVIDIA GPU clusters, managed MLOps tooling, and a serverless inference layer so they can move from experiment to production on one stack.

专项更强Offers extremely cheap serverless model hosting via API, but does not provide raw containerized GPU instances.
免费 / 开源由 PyTorch Lightning 创建团队打造的 AI 云平台

FluidStack builds and operates large-scale compute infrastructure for AI, including gigawatt-class data centers and GPU clusters used for training and inference. The company designs its own power, cooling and data center systems, onshores manufacturing and deploys solar generation to supply the power its clusters need. Rather than a self-serve cloud console, FluidStack works with AI labs and enterprises to provision large reserved clusters, with orchestration tooling for the ML frameworks and schedulers those teams already use.
FluidStack builds and operates large-scale compute infrastructure for AI, including gigawatt-class data centers and GPU clusters used for training and inference. The company designs its own power, cooling and data center systems, onshores manufacturing and deploys solar generation to supply the power its clusters need. Rather than a self-serve cloud console, FluidStack works with AI labs and enterprises to provision large reserved clusters, with orchestration tooling for the ML frameworks and schedulers those teams already use.

专项更强Scales distributed Ray workloads effortlessly across clusters, but is tailored for heavy data engineering and training.
专项更强Scales distributed Ray workloads effortlessly across clusters, but is tailored for heavy data engineering and training.
面向训练与推理工作负载的私有 AI 云