快速简洁
短短几行代码
， 即可生成线上服务
根据 Serving 服务（Predictor）模版只需要几行代码，即可完成一个自定义模型的在线推理服务逻辑。无需自己搭建复杂环境，可直接复用 OpenBayes 线上的训练环境镜像。效果 100% 复现
请求统计
内置请求统计
， 直观查看性能指标
我们的 Serving 服务可对每次请求进行跟踪，保留完整的请求结果，方便追溯问题事件。
请求指标可以对请求的性能进行分析，还可按不同周期进行筛选。清晰查看性能走向。
自定义镜像 Enterprise-Only
灵活的环境控制
， 满足企业定制化需求
私有部署可基于 Docker 创建并使用部署的自定义镜像，可更加灵活的控制部署环境，满足企业生产环境的定制化需求
多版本选择
模型版本控制
， 不再为性能对比而发愁
基于 数据仓库 的模型版本控制，可在模型的不同版本之间进行切换，直观对比推理结果
多格式支持
支持主流框架
覆盖多数应用场景
兼容主流框架 PyTorch、ONNX、XGBoost、TensorFlow 的模型格式和保存方法
%20%20--%3E%3C!DOCTYPE%20svg%20PUBLIC%20'-%2F%2FW3C%2F%2FDTD%20SVG%201.1%2F%2FEN'%20'http%3A%2F%2Fwww.w3.org%2FGraphics%2FSVG%2F1.1%2FDTD%2Fsvg11.dtd'%3E%3Csvg%20version%3D'1.1'%20id%3D'Layer_1'%20xmlns%3D'http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg'%20xmlns%3Axlink%3D'http%3A%2F%2Fwww.w3.org%2F1999%2Fxlink'%20x%3D'0px'%20y%3D'0px'%09%20viewBox%3D'0.6%20539.9%20487.3%20120.2'%20enable-background%3D'new%200.6%20539.9%20487.3%20120.2'%20xml%3Aspace%3D'preserve'%3E%3Cg%3E%09%3Cpath%20fill%3D'%23EE4C2C'%20d%3D'M63.1%2C567.3l-6.6%2C6.6c10.8%2C10.8%2C10.8%2C28.2%2C0%2C38.8c-10.8%2C10.8-28.2%2C10.8-38.8%2C0c-10.8-10.8-10.8-28.2%2C0-38.8%09%09l0%2C0l17.1-17.1l2.4-2.4l0%2C0v-12.9l-25.8%2C25.8c-14.4%2C14.4-14.4%2C37.6%2C0%2C52s37.6%2C14.4%2C51.7%2C0C77.5%2C604.8%2C77.5%2C581.7%2C63.1%2C567.3z'%2F%3E%09%3Ccircle%20fill%3D'%23EE4C2C'%20cx%3D'50.2'%20cy%3D'560.9'%20r%3D'4.8'%2F%3E%3C%2Fg%3E%3Cg%3E%09%3Cg%3E%09%09%3Cpath%20fill%3D'%23000000'%20d%3D'M129.8%2C600.3h-11.1v28.5h-8.4v-81.1c0%2C0%2C19.2%2C0%2C20.4%2C0c21.3%2C0%2C31.5%2C10.5%2C31.5%2C25.2%09%09%09C162.5%2C591%2C149.9%2C600.3%2C129.8%2C600.3z%20M130.7%2C555.8c-0.9%2C0-11.7%2C0-11.7%2C0v37.3l11.4-0.3c15.3-0.3%2C23.7-6.3%2C23.7-18.9%09%09%09C154.1%2C562.1%2C145.7%2C555.8%2C130.7%2C555.8z'%2F%3E%09%09%3Cpath%20fill%3D'%23000000'%20d%3D'M199.8%2C628.5l-4.8%2C12.9c-5.4%2C14.4-11.1%2C18.6-19.2%2C18.6c-4.5%2C0-7.8-1.2-11.4-2.7l2.4-7.5%09%09%09c2.7%2C1.5%2C5.7%2C2.7%2C9%2C2.7c4.5%2C0%2C7.8-2.4%2C12.3-13.8l3.9-10.5l-23.1-58.6h8.7l18.6%2C49l18.3-49h8.4L199.8%2C628.5z'%2F%3E%09%09%3Cpath%20fill%3D'%23000000'%20d%3D'M250.3%2C555.8v73.3h-8.4v-73.3h-28.5V548h65.2v7.8C278.5%2C555.8%2C250.3%2C555.8%2C250.3%2C555.8z'%2F%3E%09%09%3Cpath%20fill%3D'%23000000'%20d%3D'M302.3%2C630.6c-16.5%2C0-28.5-12.3-28.5-31.2c0-18.9%2C12.6-31.5%2C29.4-31.5s28.5%2C12.3%2C28.5%2C31.2%09%09%09C331.4%2C618%2C318.8%2C630.6%2C302.3%2C630.6z%20M302.6%2C575.4c-12.6%2C0-20.7%2C9.9-20.7%2C24c0%2C14.4%2C8.4%2C24.3%2C21%2C24.3s20.7-9.9%2C20.7-24%09%09%09C323.6%2C585%2C315.2%2C575.4%2C302.6%2C575.4z'%2F%3E%09%09%3Cpath%20fill%3D'%23000000'%20d%3D'M351.8%2C629.1h-8.1v-59.5l8.1-1.8v12.6c3.9-7.5%2C9.6-12.6%2C17.4-12.6c3.9%2C0%2C7.5%2C1.2%2C10.5%2C2.7l-2.1%2C7.5%09%09%09c-2.7-1.5-5.7-2.7-9-2.7c-6.3%2C0-12%2C4.8-16.8%2C15.3V629.1L351.8%2C629.1z'%2F%3E%09%09%3Cpath%20fill%3D'%23000000'%20d%3D'M411.3%2C630.6c-18%2C0-29.1-12.9-29.1-31.2c0-18.6%2C12.3-31.5%2C29.1-31.5c7.2%2C0%2C13.5%2C1.8%2C18.6%2C5.1l-2.1%2C7.2%09%09%09c-4.5-3-10.2-4.8-16.5-4.8c-12.9%2C0-20.7%2C9.6-20.7%2C23.7c0%2C14.4%2C8.4%2C24%2C21%2C24c6%2C0%2C12-1.8%2C16.5-4.8l1.8%2C7.5%09%09%09C424.5%2C628.8%2C418.2%2C630.6%2C411.3%2C630.6z'%2F%3E%09%09%3Cpath%20fill%3D'%23000000'%20d%3D'M479.5%2C629.1v-38.5c0-10.5-4.2-15-12.6-15c-6.9%2C0-13.5%2C3.6-18.3%2C8.4v45.1h-8.1v-87.4l8.1-1.8%09%09%09c0%2C0%2C0%2C37.3%2C0%2C37.6c6.3-6.3%2C14.1-9.3%2C20.7-9.3c11.4%2C0%2C18.6%2C7.5%2C18.6%2C20.4v40.6H479.5z'%2F%3E%09%3C%2Fg%3E%3C%2Fg%3E%3C%2Fsvg%3E "PyTorch")
'%3E%3ClinearGradient%20id%3D'SVGID_3_'%20gradientUnits%3D'userSpaceOnUse'%20x1%3D'-1.6'%20y1%3D'335.05'%20x2%3D'53.6'%20y2%3D'335.05'%20gradientTransform%3D'translate(0%20-304)'%3E%3Cstop%20offset%3D'0'%20stop-color%3D'%23ff6f00'%2F%3E%3Cstop%20offset%3D'1'%20stop-color%3D'%23ffa800'%2F%3E%3C%2FlinearGradient%3E%3Cpath%20d%3D'M-1.6%204.6h55.2v52.9H-1.6V4.6z'%20fill%3D'url(%23SVGID_3_)'%2F%3E%3C%2Fg%3E%3C%2Fg%3E%3C%2Fg%3E%3Cg%3E%3Cg%3E%3Cdefs%3E%3Cpath%20id%3D'SVGID_4_'%20d%3D'M.5%2017.6L23%204.8v52.6l-9-5.2V21.4L.5%2029.3z'%2F%3E%3C%2Fdefs%3E%3CclipPath%20id%3D'SVGID_5_'%3E%3Cuse%20xlink%3Ahref%3D'%23SVGID_4_'%20overflow%3D'visible'%2F%3E%3C%2FclipPath%3E%3Cg%20clip-path%3D'url(%23SVGID_5_)'%3E%3ClinearGradient%20id%3D'SVGID_6_'%20gradientUnits%3D'userSpaceOnUse'%20x1%3D'-1.9'%20y1%3D'335.05'%20x2%3D'53.3'%20y2%3D'335.05'%20gradientTransform%3D'translate(0%20-304)'%3E%3Cstop%20offset%3D'0'%20stop-color%3D'%23ff6f00'%2F%3E%3Cstop%20offset%3D'1'%20stop-color%3D'%23ffa800'%2F%3E%3C%2FlinearGradient%3E%3Cpath%20d%3D'M-1.9%204.6h55.2v52.9H-1.9V4.6z'%20fill%3D'url(%23SVGID_6_)'%2F%3E%3C%2Fg%3E%3C%2Fg%3E%3C%2Fg%3E%3Cpath%20style%3D'fill%3A%23425066'%20d%3D'M88.2%2021.1h-10v27.7h-5.6V21.1h-10v-4.5h25.6v4.5z'%2F%3E%3Cpath%20style%3D'fill%3A%23425066'%20d%3D'M94.9%2049.2c-3.4%200-6.2-1.1-8.3-3.2-2.1-2.1-3.2-5-3.2-8.6v-.7c0-2.2.4-4.4%201.4-6.4.9-1.8%202.2-3.3%203.9-4.4%201.7-1.1%203.6-1.6%205.6-1.6%203.3%200%205.8%201%207.6%203.1s2.7%205%202.7%208.8v2.2H88.9c.1%201.8.8%203.4%202%204.7%201.2%201.2%202.7%201.8%204.4%201.7%202.4.1%204.6-1.1%206-3l2.9%202.8c-1%201.4-2.3%202.6-3.8%203.3-1.8%201-3.6%201.4-5.5%201.3zm-.6-20.5c-1.4-.1-2.7.5-3.6%201.5-1%201.2-1.6%202.7-1.7%204.3h10.3v-.4c-.1-1.8-.6-3.2-1.4-4.1-1-.8-2.3-1.4-3.6-1.3zm19.4-3.9l.2%202.8c1.7-2.1%204.3-3.3%207-3.2%205%200%207.5%202.9%207.6%208.6v15.8h-5.4V33.3c0-1.5-.3-2.6-1-3.4-.7-.7-1.7-1.1-3.2-1.1-2.1-.1-4%201.1-4.9%202.9v17h-5.4v-24l5.1.1zm32.2%2017.5c0-.9-.4-1.7-1.2-2.2-1.2-.7-2.6-1.1-3.9-1.3-1.6-.3-3.1-.8-4.6-1.5-2.7-1.3-4-3.2-4-5.6%200-2%201-4%202.6-5.2%201.7-1.4%204-2.1%206.6-2.1%202.9%200%205.2.7%206.9%202.1%201.7%201.3%202.7%203.4%202.6%205.5h-5.4c0-1-.4-1.9-1.2-2.6-.9-.7-1.9-1.1-3.1-1-1%200-2%20.2-2.9.8-.7.5-1.1%201.3-1.1%202.2%200%20.8.4%201.5%201%201.9.7.5%202.1.9%204.2%201.4%201.7.3%203.4.9%205%201.7%201.1.5%202%201.3%202.7%202.3.6%201%20.9%202.1.9%203.3%200%202.1-1%204-2.7%205.2-1.8%201.3-4.1%202-7%202-1.8%200-3.6-.3-5.2-1.1-1.4-.6-2.7-1.6-3.6-2.9-.8-1.2-1.3-2.6-1.3-4h5.2c0%201.1.5%202.2%201.4%202.9%201%20.7%202.3%201.1%203.5%201%201.4%200%202.5-.3%203.2-.8%201-.4%201.4-1.2%201.4-2zm8.1-5.7c0-2.2.4-4.4%201.4-6.3.9-1.8%202.2-3.3%203.9-4.3%201.8-1%203.8-1.6%205.8-1.5%203.2%200%205.9%201%207.9%203.1s3.1%204.8%203.3%208.3v1.3c0%202.2-.4%204.3-1.4%206.3-.8%201.8-2.2%203.3-3.9%204.3-1.8%201-3.8%201.6-5.9%201.5-3.4%200-6.1-1.1-8.1-3.4-2-2.2-3-5.2-3.1-9l.1-.3zm5.3.5c0%202.5.5%204.4%201.5%205.8%201.8%202.3%205.1%202.8%207.5%201%20.4-.3.7-.6%201-1%201-1.4%201.5-3.5%201.5-6.2%200-2.4-.5-4.3-1.6-5.8-1.7-2.3-5-2.8-7.4-1.1-.4.3-.8.7-1.1%201-.8%201.4-1.4%203.5-1.4%206.3zm33.1-7.3c-.7-.1-1.5-.2-2.2-.2-2.5%200-4.1.9-5%202.8v16.4h-5.4v-24h5.1l.1%202.7c1.3-2.1%203.1-3.1%205.4-3.1.6%200%201.3.1%201.9.3l.1%205.1zm22.5%205.3h-13v13.7h-5.6V16.6h20.5v4.5h-14.9v9.6h13v4.4zm10.7%2013.7h-5.4V16.5h5.4v32.3zm3.9-12.2c0-2.2.4-4.4%201.4-6.3.9-1.8%202.2-3.3%203.9-4.3%201.8-1%203.8-1.6%205.8-1.5%203.2%200%205.9%201%207.9%203.1s3.1%204.8%203.3%208.3v1.3c0%202.2-.4%204.3-1.3%206.3-.8%201.8-2.2%203.3-3.9%204.3-1.8%201-3.8%201.6-5.9%201.5-3.4%200-6.1-1.1-8.1-3.4-2-2.2-3.1-5.2-3.1-9v-.3zm5.4.5c0%202.5.5%204.4%201.5%205.8%201%201.4%202.6%202.2%204.3%202.1%201.7.1%203.3-.7%204.2-2.1%201-1.4%201.5-3.5%201.5-6.2%200-2.4-.5-4.3-1.6-5.8-1.7-2.3-5-2.8-7.4-1.1-.4.3-.8.7-1.1%201-.9%201.4-1.4%203.5-1.4%206.3zm41.2%204.3l3.8-16.5h5.2l-6.5%2024h-4.4l-5.1-16.5-5.1%2016.5h-4.4l-6.6-24h5.3l3.9%2016.4%204.9-16.4h4.1l4.9%2016.5z'%2F%3E%3C%2Fsvg%3E "TensorFlow")
在线演示
根据不同场景和算力，
提供对应优化的模型推理
JSON 输出
请求耗时 … 毫秒
Loading…
JSON 输出
请求耗时 … 毫秒
Loading…JSON 输出
请求耗时 … 毫秒
Loading…
JSON 输出
请求耗时 … 毫秒
Loading…高度灵活
支持 OpenBayes 标准的
predictor.py 方式部署。针对高级用户，也可以完全自定义，绕过 OpenBayes 提供的框架，手动编写自定义部署服务。多框架支持
兼容主流框架的模型格式。支持 PyTorch、XGBoost、TensorFlow 的多种模型文件保存方法
GPU 分片 Enterprise-Only
私有部署可将物理 GPU 划分为多个虚拟 GPU，并保证计算资源与显存资源的隔离，大幅提升 GPU 利用效率，降低企业采购成本
平台定价方案
从入门级 CPU，
到专业级高性能 GPU 算力
我们提供了多种类型的算力方案，可满足不同算力需求、不同预算的用户
NVIDIA Tesla T4
2.99元/小时起
适合初次尝试机器学习及对算力要求较低的用户
免费试用 - 1x NVIDIA T4
- 16 GB 显存
- 4 核 vCPU
- 30 GB 内存
- 50 GB 工作空间
- 65 TFLOPS 算力
NVIDIA Tesla V100
11.24元/小时起
适合对性能以及算力有一定需求的高级用户
免费试用 - 1x NV Tesla V100
- 16 GB 显存
- 4 核 vCPU
- 40 GB 内存
- 50 GB 工作空间
- 100 TFLOPS 算力
私有部署
按需定价
适合对硬件以及功能有定制需求的企业级用户
- 4x-8x NVLink
- 64+ GB 显存
- 12+ 多核 CPU
- 128+ GB 内存
- 20+ TB 存储空间
- 1+ PFLOPS 算力