由 DeepSeek-V4 与 R1 驱动的高效开源权重与托管推理 AI,token 成本极低。
覆盖 459 类工作流,通过流量估算、产品功能与替代方案辅助选型。
“Python Implementation” · 基础模型与底层架构·140 个结果
重新开始由 DeepSeek-V4 与 R1 驱动的高效开源权重与托管推理 AI,token 成本极低。
面向 AI 与数据负载的无服务器云基础设施



一个兼容 OpenAI 的 API,接入 500+ 图像、视频、音频与大模型

Metaflow is Netflix's open-source Python framework for building, running, and operating real-world machine learning, AI, and data science workflows. It keeps your code in plain Python while handling dependency management, versioning, orchestration, and elastic cloud compute.
Metaflow is Netflix's open-source Python framework for building, running, and operating real-world machine learning, AI, and data science workflows. It keeps your code in plain Python while handling dependency management, versioning, orchestration, and elastic cloud compute.

不受限、兼容 OpenAI 的 LLM API,由你自己的策略治理
不受限、兼容 OpenAI 的 LLM API,由你自己的策略治理
用一个 API 快速、可预测成本地运行大语言模型与多模态模型


一个通用 API,接入 500+ AI 模型与 1,000+ 智能体工具
Abyss turns a Python project into a shareable, pay-per-run web tool. You upload your scripts, describe the inputs, and Abyss wraps them in a Widget with a form, a public link, and a price per run — while your API keys stay on your side.

在一个订阅中并排对比 ChatGPT、Claude、Gemini 及 30 多款 AI 模型


面向编程智能体的智能模型路由

专业 AI 模型托管、优化端点与按需 GPU 基础设施
Open Interpreter is an open-source coding agent that lets language models write and run code directly on your own machine, turning plain-English instructions into real files, charts, and automations.
Open Interpreter is an open-source coding agent that lets language models write and run code directly on your own machine, turning plain-English instructions into real files, charts, and automations.



为智能体快速提供开源模型推理,附带热缓存路由
为智能体快速提供开源模型推理,附带热缓存路由
General Compute is an inference cloud built around alternative AI silicon rather than standard GPU fleets. The company buys racks of accelerators, racks them in colocation data centres, and handles the cooling, power, and networking itself. On top of that hardware it runs a disaggregated inference stack that splits each request: prompt processing runs on NVIDIA GPUs for prefill, while token generation runs on faster decode chips such as Cerebras, SambaNova, Positron AI, and d-Matrix. Customers get bare metal access or a tokens-per-second SLA.
General Compute is an inference cloud built around alternative AI silicon rather than standard GPU fleets. The company buys racks of accelerators, racks them in colocation data centres, and handles the cooling, power, and networking itself. On top of that hardware it runs a disaggregated inference stack that splits each request: prompt processing runs on NVIDIA GPUs for prefill, while token generation runs on faster decode chips such as Cerebras, SambaNova, Positron AI, and d-Matrix. Customers get bare metal access or a tokens-per-second SLA.

高吞吐开源权重模型推理,按延迟分档定价
看看霍格沃茨哪个学院会选择你