HackerNews中文版

我们刚刚发布了开放大语言模型规范（Open LLM Specification，简称 OLLS）——这是一个由社区驱动的标准，旨在统一开发者与大型语言模型（LLM）之间的交互方式，适用于 OpenAI、Anthropic、Google 等多个提供商。目前，每个提供商的请求/响应格式各不相同，这使得集成变得困难： - 解析响应不一致 - 切换模型需要自定义包装 - 错误处理和元数据差异巨大 OLLS 定义了一种简单、可扩展的 JSON 规范，适用于输入（提示、参数、元数据）和输出（内容、推理、使用情况、错误）。可以将其视为 LLM 的 OpenAPI——可移植、可预测且与提供商无关。 GitHub 仓库 - [https://github.com/julurisaichandu/open-llm-specification](https://github.com/julurisaichandu/open-llm-specification) 示例输入/输出格式、目标和路线图我们正在寻找贡献者、反馈和实际应用案例！让我们共同构建一个统一的 LLM 接口——欢迎贡献想法或加入讨论。

查看原文

We have just launched the Open LLM Specification (OLLS) – a community-driven standard that unifies how developers interact with large language models (LLMs) across providers such as OpenAI, Anthropic, Google, and others.Right now, every provider has different request/response formats, which makes integration painful:Parsing responses is inconsistentSwitching models needs custom wrappersError handling and metadata vary wildlyOLLS defines a simple, extensible JSON spec for both inputs (prompts, parameters, metadata) and outputs (content, reasoning, usage, errors). Think of it like OpenAPI for LLMs—portable, predictable, and provider-agnostic.GitHub Repo - <a href="https://github.com/julurisaichandu/open-llm-specification">https://github.com/julurisaichandu/open-llm-specification</a> Example input/output formats, goals, and roadmap Looking for contributors, feedback, and real-world use cases!Let’s build a unified LLM interface—contribute ideas or join the discussion

展示HN：OLL – 一种与供应商无关的LLM输入和输出标准