{ "slug": "ovhcloud-ai-endpoints", "name": "OVHcloud AI Endpoints", "company": "OVH Groupe SA", "website": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/", "category": "platforms", "business_model": "freemium", "availability": "global", "launched": "2025", "founded": "1999", "hq_country": "FR", "docs_url": "https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-responses-api", "pricing_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/catalog", "tags": [ "cloud", "inference", "serverless", "open-models", "europe" ], "description": "Serverless inference API from OVHcloud that serves open-weight language, vision, embedding, image and speech models through OpenAI-compatible endpoints, billed by usage.", "overview": "OVHcloud AI Endpoints is a serverless inference service operated by OVH Groupe SA, the French cloud provider founded in 1999 and headquartered in Roubaix. It gives developers OpenAI-compatible API access to more than 40 open-weight models — Llama, Qwen, Mistral, gpt-oss, Whisper and Stable Diffusion among them — covering chat, reasoning, vision, embeddings, image generation, transcription and moderation without requiring users to deploy GPUs. OVHcloud launched the service on 29 April 2025 and bills it pay-as-you-go per million tokens, with a US$200 Public Cloud trial credit for new projects. The limitation is on the newest route: OVHcloud's own documentation says the OpenAI-compatible /v1/responses endpoint has no statefulness and no built-in tools such as web search, so callers must send store: false and rely on custom function tools.", "features": [ "Serverless, OpenAI-compatible inference: the /v1/chat/completions and /v1/responses routes", "More than 40 models across LLMs, visual LLMs, embeddings, image generation, speech-to-text and text-to-speech", "Virtual models that map a request to the best available model by size, price or capability", "Zero data retention: OVHcloud states customer data is never used to train or improve its models", "Per-model pricing published in the catalogue, in euros per million tokens", "Model lifecycle management with transparent versioning, hosted on OVHcloud's own European infrastructure" ], "use_cases": [ "Adding chat, reasoning or coding assistance to an application without managing GPUs", "Getting embeddings, images or speech from the same OpenAI-compatible key", "Keeping inference inside European infrastructure for data-residency requirements" ], "pricing_notes": "Billed by usage rather than by subscription: each model's price per million tokens is published in the public catalogue, ranging from around €0.01 for an embedding model to about €0.91 for a large visual model, with a few models (moderation and image generation) listed as free. OVHcloud offers a US$200 Public Cloud free-trial credit to first-time projects. Rates change, so the catalogue is the source of record; this record sets no comparable paid_from figure.", "faq": [ { "q": "What is OVHcloud AI Endpoints?", "a": "A serverless inference API: it exposes pre-trained open-weight models for chat, reasoning, vision, embeddings, image generation and speech over OpenAI-compatible HTTP routes, so an application does not need to run its own GPUs." }, { "q": "Where is OVHcloud based?", "a": "In Roubaix, France. OVHcloud's about page calls the company \"a French company\" founded in 1999, and the Public Cloud terms name OVH Groupe SA at 2 rue Kellermann, 59100 Roubaix." }, { "q": "How is it billed?", "a": "Pay-as-you-go per million tokens, with each model's rate published in the catalogue, plus a US$200 free-trial credit for first-time Public Cloud projects. No subscription is required." }, { "q": "What are the limits of the API?", "a": "On the /v1/responses route, OVHcloud's documentation says statefulness is not managed (always send store: false) and built-in tools such as web search, file search or code execution are unsupported; only custom function tools work." }, { "q": "What does this record not establish?", "a": "No page read for this record states the product's own interface languages, its licence terms, or a per-region service list, so those fields stay unconfirmed." } ], "i18n": { "zh": { "description": "OVHcloud 提供的无服务器推理 API，通过兼容 OpenAI 的端点提供开放权重语言、视觉、嵌入、图像与语音模型，按用量计费。", "overview": "OVHcloud AI Endpoints 是由 OVH Groupe SA 运营的无服务器推理服务；该公司是 1999 年成立、总部位于法国鲁贝的云服务商。它向开发者提供兼容 OpenAI 的 API，可调用 40 多个开放权重模型——包括 Llama、Qwen、Mistral、gpt-oss、Whisper 与 Stable Diffusion——覆盖对话、推理、视觉、嵌入、图像生成、转写与内容审核，而无需自行部署 GPU。OVHcloud 于 2025 年 4 月 29 日发布该服务，按每百万 Token 用量计费，新项目可获 200 美元 Public Cloud 试用额度。局限出现在最新的路由上：据 OVHcloud 自家文档，兼容 OpenAI 的 /v1/responses 端点不支持状态保持，也不支持网页搜索等内置工具，调用方必须传 store: false 并只能使用自定义函数工具。", "features": [ "无服务器、兼容 OpenAI 的推理：/v1/chat/completions 与 /v1/responses 路由", "40 多个模型，涵盖 LLM、视觉 LLM、嵌入、图像生成、语音转文字与文字转语音", "虚拟模型：按规模、价格或能力把请求映射到最合适的可用模型", "零数据保留：OVHcloud 表示客户数据不会用于训练或改进其模型", "每个模型的价格在模型目录中公布，以欧元/百万 Token 计", "模型生命周期管理，版本公开透明，运行在 OVHcloud 自有欧洲基础设施上" ], "use_cases": [ "无需管理 GPU 就为应用加入对话、推理或编程辅助能力", "用同一个兼容 OpenAI 的密钥获取嵌入、图像或语音能力", "为满足数据驻留要求，把推理留在欧洲基础设施内" ], "pricing_notes": "按用量而非订阅计费：每个模型每百万 Token 的价格在公开目录中列出，从嵌入模型的约 0.01 欧元到大视觉模型的约 0.91 欧元不等，另有少数模型（审核与图像生成）标注为免费。OVHcloud 为首次创建 Public Cloud 项目提供 200 美元试用额度。费率会变化，因此以模型目录为准；本记录不设可比的 paid_from 价格。", "faq": [ { "q": "OVHcloud AI Endpoints 是什么？", "a": "一套无服务器推理 API：通过兼容 OpenAI 的 HTTP 路由提供预训练的开放权重模型，覆盖对话、推理、视觉、嵌入、图像生成与语音，让应用无需自建 GPU。" }, { "q": "OVHcloud 总部在哪里？", "a": "法国鲁贝。OVHcloud 的关于页称其为 1999 年成立的“法国公司”，Public Cloud 条款中列明 OVH Groupe SA，地址为 2 rue Kellermann, 59100 Roubaix。" }, { "q": "如何计费？", "a": "按每百万 Token 用量计费，每个模型的费率在目录中公布；首次使用 Public Cloud 的项目可获 200 美元试用额度。无需订阅。" }, { "q": "该 API 有哪些限制？", "a": "据 OVHcloud 文档，/v1/responses 路由不管理状态（必须传 store: false），且不支持网页搜索、文件检索、代码执行等内置工具，只能使用自定义函数工具。" }, { "q": "这条记录没有确立哪些内容？", "a": "本次读取的页面均未写明产品自身的界面语言、许可条款或按地区的可用范围，因此这些字段仍未确认。" } ], "facts": { "ovhcloud-ai-endpoints.hq.france": { "conditions": "据 ovhcloud.com/en/about-us/ 于 2026-10-12 所述", "note": "在厂商“关于”页面上匹配到: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, "ovhcloud-ai-endpoints.pricing.usage-and-trial": { "value": "按百万 token 计费、按模型公布费率；赠送 US$200 公有云试用额度", "conditions": "据 ovhcloud.com 于 2026-10-12 所述", "note": "产品页面提供\"US$ 200 in free credit\"，目录以 €/Mtoken 列出各模型价格（例如 embedding 0.01 €、大型视觉模型 0.91 €），少数模型显示为\"Free\"。未公布任何周期性方案价格。" }, "ovhcloud-ai-endpoints.api.openai-routes": { "value": "兼容 OpenAI 的 /v1/chat/completions 与 /v1/responses 路由", "conditions": "据 docs.ovhcloud.com 于 2026-10-12 的文档所述", "note": "文档将 /v1/responses 称为\"the most recent OpenAI-compatible route\"，并说明其与 v1/chat/completions 一样可用于文本生成、多轮对话和工具调用。" } }, "selection_evidence": { "tasks": { "note": "从记录推导：类别“platforms”。" }, "form_factor": { "note": "在厂商产品页面上匹配到：\"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "note": "由记录描述/功能/标签中的平台用词推导而来。" }, "deployment": { "note": "厂商称该服务运行“on OVHcloud's robust and confidential infrastructure”；它通过网络访问，而非安装在用户机器上。" }, "offline": { "note": "记录中不包含离线运行的声明。" }, "regions": { "note": "与记录中已有的可用地域字段（“global”）一致；厂商称服务 140 个国家，且未读取按区域划分的服务清单。" }, "api": { "note": "厂商文档记录了 OpenAI 兼容的 HTTP 路由，包括 /v1/chat/completions 与 /v1/responses。" }, "license": { "note": "商业产品；记录中未发布源代码许可证。" }, "commercial_use": { "note": "商业产品；所读页面未声明非商业限制，但这一点未经验证。" }, "billing": { "note": "厂商发布了按模型的用量费率与 US$200 试用额度，但没有周期性方案价格，因此未设置可比 paid_from 数值。" }, "languages": { "note": "本轮所读的任何厂商页面都未说明产品自身的界面语言；保留该键是为了日后能作答，而不是留空。" } } }, "es": { "description": "API de inferencia serverless de OVHcloud que sirve modelos abiertos de lenguaje, visión, embeddings, imagen y voz mediante endpoints compatibles con OpenAI y facturación por uso.", "overview": "OVHcloud AI Endpoints es un servicio de inferencia serverless operado por OVH Groupe SA, el proveedor de nube francés fundado en 1999 y con sede en Roubaix. Ofrece a los desarrolladores acceso por API compatible con OpenAI a más de 40 modelos de pesos abiertos —entre ellos Llama, Qwen, Mistral, gpt-oss, Whisper y Stable Diffusion— que cubren chat, razonamiento, visión, embeddings, generación de imágenes, transcripción y moderación sin desplegar GPU. OVHcloud lanzó el servicio el 29 de abril de 2025 y lo factura por uso por millón de tokens, con un crédito de prueba de 200 US$ para proyectos nuevos de Public Cloud. La limitación está en la ruta más reciente: la documentación del propio OVHcloud indica que el endpoint /v1/responses compatible con OpenAI no admite estado ni herramientas integradas como la búsqueda web, por lo que hay que enviar store: false y usar solo funciones personalizadas.", "features": [ "Inferencia serverless compatible con OpenAI: rutas /v1/chat/completions y /v1/responses", "Más de 40 modelos: LLM, LLM visuales, embeddings, generación de imágenes, voz a texto y texto a voz", "Modelos virtuales que asignan cada solicitud al mejor modelo disponible por tamaño, precio o capacidad", "Retención cero: OVHcloud afirma que los datos de los clientes nunca se usan para entrenar o mejorar sus modelos", "Precio por modelo publicado en el catálogo, en euros por millón de tokens", "Gestión del ciclo de vida de los modelos con versionado transparente, sobre la infraestructura europea de OVHcloud" ], "use_cases": [ "Añadir chat, razonamiento o asistencia de código a una aplicación sin gestionar GPU", "Obtener embeddings, imágenes o voz con la misma clave compatible con OpenAI", "Mantener la inferencia dentro de infraestructura europea por requisitos de residencia de datos" ], "pricing_notes": "Se factura por uso y no por suscripción: el precio por millón de tokens de cada modelo figura en el catálogo público, desde unos 0,01 € para un modelo de embeddings hasta unos 0,91 € para un modelo visual grande, con algunos modelos (moderación y generación de imágenes) marcados como gratuitos. OVHcloud ofrece un crédito de prueba de 200 US$ para primeros proyectos de Public Cloud. Las tarifas cambian, así que el catálogo es la referencia; este registro no fija ninguna cifra paid_from comparable.", "faq": [ { "q": "¿Qué es OVHcloud AI Endpoints?", "a": "Una API de inferencia serverless: expone modelos abiertos preentrenados para chat, razonamiento, visión, embeddings, generación de imágenes y voz mediante rutas HTTP compatibles con OpenAI, de modo que la aplicación no necesita ejecutar sus propias GPU." }, { "q": "¿Dónde está radicada OVHcloud?", "a": "En Roubaix, Francia. Su página corporativa la describe como \"una empresa francesa\" fundada en 1999, y las condiciones de Public Cloud nombran a OVH Groupe SA, 2 rue Kellermann, 59100 Roubaix." }, { "q": "¿Cómo se factura?", "a": "Por uso, por millón de tokens, con la tarifa de cada modelo publicada en el catálogo, más un crédito de prueba de 200 US$ para los primeros proyectos de Public Cloud. No requiere suscripción." }, { "q": "¿Cuáles son los límites de la API?", "a": "Según la documentación de OVHcloud, la ruta /v1/responses no gestiona estado (hay que enviar store: false) y no admite herramientas integradas como búsqueda web, búsqueda de archivos o ejecución de código; solo funcionan las funciones personalizadas." }, { "q": "¿Qué no establece este registro?", "a": "Ninguna página leída indica los idiomas de interfaz del producto, sus términos de licencia ni una lista de disponibilidad por región, así que esos campos siguen sin confirmar." } ], "facts": { "ovhcloud-ai-endpoints.hq.france": { "conditions": "Según lo indicado en ovhcloud.com/en/about-us/ el 2026-10-12", "note": "Coincidencia en la página \"Acerca de\" del proveedor: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, "ovhcloud-ai-endpoints.pricing.usage-and-trial": { "value": "Pago por uso por millón de tokens, con tarifas publicadas por modelo; crédito de prueba de US$200 en la nube pública", "conditions": "Según lo indicado en ovhcloud.com el 2026-10-12", "note": "La página del producto ofrece \"US$ 200 in free credit\" y el catálogo lista precios por modelo en €/Mtoken (p. ej. embeddings 0.01 €, un modelo visual grande 0.91 €), con algunos modelos mostrados como \"Free\". No se publica ningún precio de plan periódico." }, "ovhcloud-ai-endpoints.api.openai-routes": { "value": "Rutas /v1/chat/completions y /v1/responses compatibles con OpenAI", "conditions": "Según lo documentado en docs.ovhcloud.com el 2026-10-12", "note": "La documentación describe /v1/responses como \"the most recent OpenAI-compatible route\" y, al igual que v1/chat/completions, utilizable para generación de texto, conversaciones multiturno y llamadas a herramientas." } }, "selection_evidence": { "tasks": { "note": "Derivado del registro: categoría «platforms»." }, "form_factor": { "note": "Encontrado en la página de producto del proveedor: \"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "note": "Derivado de las palabras de plataforma en la descripción, las funciones y las etiquetas del registro." }, "deployment": { "note": "El proveedor afirma que el servicio funciona «on OVHcloud's robust and confidential infrastructure»; se accede por red, no se instala en la máquina del usuario." }, "offline": { "note": "El registro no contiene ninguna afirmación de funcionamiento sin conexión." }, "regions": { "note": "Refleja el campo de disponibilidad ya presente en el registro («global»); el proveedor dice que presta servicio en 140 países y no se leyó ninguna lista de servicios por región." }, "api": { "note": "La documentación del proveedor documenta rutas HTTP compatibles con OpenAI, incluidas /v1/chat/completions y /v1/responses." }, "license": { "note": "Producto comercial; no se publica licencia del código fuente en el registro." }, "commercial_use": { "note": "Producto comercial; las páginas consultadas no indican una restricción no comercial, pero eso no se verificó." }, "billing": { "note": "El proveedor publica tarifas de uso por modelo y un crédito de prueba de US$200, pero ningún precio de plan periódico, por lo que no se fija una cifra paid_from comparable." }, "languages": { "note": "Ninguna página del proveedor consultada en esta pasada indica los idiomas de interfaz del propio producto; la clave existe para poder responder al campo más adelante en lugar de dejarlo ausente." } } }, "fr": { "description": "API d'inférence serverless d'OVHcloud qui sert des modèles ouverts de langage, vision, embeddings, image et parole via des points de terminaison compatibles OpenAI, facturée à l'usage.", "overview": "OVHcloud AI Endpoints est un service d'inférence serverless exploité par OVH Groupe SA, fournisseur cloud français fondé en 1999 et basé à Roubaix. Il donne aux développeurs un accès API compatible OpenAI à plus de 40 modèles à poids ouverts — dont Llama, Qwen, Mistral, gpt-oss, Whisper et Stable Diffusion — couvrant le chat, le raisonnement, la vision, les embeddings, la génération d'images, la transcription et la modération, sans déployer de GPU. OVHcloud a lancé le service le 29 avril 2025 et le facture à l'usage par million de tokens, avec un crédit d'essai de 200 US$ pour les nouveaux projets Public Cloud. La limite concerne la route la plus récente : la documentation d'OVHcloud indique que le point de terminaison /v1/responses compatible OpenAI ne gère ni l'état ni les outils intégrés comme la recherche web ; il faut envoyer store: false et n'utiliser que des fonctions personnalisées.", "features": [ "Inférence serverless compatible OpenAI : routes /v1/chat/completions et /v1/responses", "Plus de 40 modèles : LLM, LLM visuels, embeddings, génération d'images, parole vers texte et texte vers parole", "Modèles virtuels qui orientent chaque requête vers le meilleur modèle disponible selon la taille, le prix ou les capacités", "Rétention nulle : OVHcloud déclare que les données clients ne servent jamais à entraîner ou améliorer ses modèles", "Prix par modèle publié dans le catalogue, en euros par million de tokens", "Gestion du cycle de vie des modèles avec versionnage transparent, sur l'infrastructure européenne d'OVHcloud" ], "use_cases": [ "Ajouter chat, raisonnement ou assistance au code à une application sans gérer de GPU", "Obtenir embeddings, images ou parole avec la même clé compatible OpenAI", "Garder l'inférence dans une infrastructure européenne pour des exigences de résidence des données" ], "pricing_notes": "Facturation à l'usage et non par abonnement : le prix par million de tokens de chaque modèle est publié dans le catalogue public, d'environ 0,01 € pour un modèle d'embeddings à environ 0,91 € pour un grand modèle visuel, quelques modèles (modération et génération d'images) étant indiqués comme gratuits. OVHcloud offre un crédit d'essai de 200 US$ aux premiers projets Public Cloud. Les tarifs évoluent : le catalogue fait référence ; cette fiche ne fixe aucun montant paid_from comparable.", "faq": [ { "q": "Qu'est-ce qu'OVHcloud AI Endpoints ?", "a": "Une API d'inférence serverless : elle expose des modèles ouverts pré-entraînés pour le chat, le raisonnement, la vision, les embeddings, la génération d'images et la parole via des routes HTTP compatibles OpenAI, sans que l'application ait à exploiter ses propres GPU." }, { "q": "Où OVHcloud est-elle établie ?", "a": "À Roubaix, en France. Sa page institutionnelle la présente comme « une entreprise française » fondée en 1999, et les conditions Public Cloud nomment OVH Groupe SA, 2 rue Kellermann, 59100 Roubaix." }, { "q": "Comment est-ce facturé ?", "a": "À l'usage, par million de tokens, avec le tarif de chaque modèle publié dans le catalogue, plus un crédit d'essai de 200 US$ pour les premiers projets Public Cloud. Aucun abonnement n'est requis." }, { "q": "Quelles sont les limites de l'API ?", "a": "Selon la documentation d'OVHcloud, la route /v1/responses ne gère pas l'état (envoyez toujours store: false) et ne prend pas en charge les outils intégrés comme la recherche web, la recherche de fichiers ou l'exécution de code ; seules les fonctions personnalisées fonctionnent." }, { "q": "Que n'établit pas cette fiche ?", "a": "Aucune page consultée n'indique les langues d'interface du produit, ses termes de licence ni une liste de disponibilité par région ; ces champs restent donc non confirmés." } ], "facts": { "ovhcloud-ai-endpoints.hq.france": { "conditions": "Comme indiqué sur ovhcloud.com/en/about-us/ le 2026-10-12", "note": "Correspondance sur la page « À propos » de l’éditeur: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, "ovhcloud-ai-endpoints.pricing.usage-and-trial": { "value": "Paiement à l'usage par million de tokens, tarifs publiés par modèle ; crédit d'essai de 200 $US sur le cloud public", "conditions": "Comme indiqué sur ovhcloud.com le 2026-10-12", "note": "La page produit offre « US$ 200 in free credit » et le catalogue liste les prix par modèle en €/Mtoken (p. ex. embeddings 0.01 €, un grand modèle visuel 0.91 €), quelques modèles étant affichés comme « Free ». Aucun prix de formule périodique n'est publié." }, "ovhcloud-ai-endpoints.api.openai-routes": { "value": "Routes /v1/chat/completions et /v1/responses compatibles avec OpenAI", "conditions": "Comme documenté sur docs.ovhcloud.com le 2026-10-12", "note": "La documentation décrit /v1/responses comme « the most recent OpenAI-compatible route » et, comme v1/chat/completions, utilisable pour la génération de texte, les conversations multi-tours et l'appel d'outils." } }, "selection_evidence": { "tasks": { "note": "Déduit de la fiche : catégorie « platforms »." }, "form_factor": { "note": "Correspondance sur la page produit du fournisseur : \"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "note": "Déduit des mots de plateforme présents dans la description, les fonctionnalités et les étiquettes de la fiche." }, "deployment": { "note": "Le fournisseur indique que le service tourne « on OVHcloud's robust and confidential infrastructure » ; il est accessible via le réseau et n'est pas installé sur la machine de l'utilisateur." }, "offline": { "note": "La fiche ne contient aucune affirmation de fonctionnement hors ligne." }, "regions": { "note": "Reprend le champ de disponibilité déjà présent dans la fiche (« global ») ; le fournisseur indique desservir 140 pays, et aucune liste de services par région n'a été lue." }, "api": { "note": "La documentation du fournisseur documente des routes HTTP compatibles OpenAI, dont /v1/chat/completions et /v1/responses." }, "license": { "note": "Produit commercial ; aucune licence du code source n'est publiée dans la fiche." }, "commercial_use": { "note": "Produit commercial ; les pages consultées n'indiquent pas de restriction non commerciale, mais cela n'a pas été vérifié." }, "billing": { "note": "Le fournisseur publie des tarifs d'usage par modèle et un crédit d'essai de US$200, mais aucun prix de forfait périodique ; aucune valeur paid_from comparable n'est donc définie." }, "languages": { "note": "Aucune page du fournisseur consultée lors de cette passe n'indique les langues d'interface du produit lui-même ; la clé existe pour que le champ puisse être renseigné plus tard plutôt que laissé absent." } } }, "ja": { "description": "OVHcloud のサーバーレス推論 API。OpenAI 互換エンドポイント経由でオープンウェイトの言語・視覚・埋め込み・画像・音声モデルを提供し、従量課金で利用する。", "overview": "OVHcloud AI Endpoints は、1999 年創業でフランス・ルーベに本社を置くクラウド事業者 OVH Groupe SA が運営するサーバーレス推論サービスです。開発者向けに OpenAI 互換 API を提供し、Llama、Qwen、Mistral、gpt-oss、Whisper、Stable Diffusion など 40 以上のオープンウェイトモデルを、GPU を用意せずにチャット、推論、視覚、埋め込み、画像生成、文字起こし、モデレーションに利用できます。OVHcloud は 2025 年 4 月 29 日に本サービスを発表し、100 万トークン単位の従量課金とし、新規 Public Cloud プロジェクトには 200 米ドルのトライアルクレジットを付与します。制約は最新のルートにあります。OVHcloud のドキュメントによれば、OpenAI 互換の /v1/responses エンドポイントは状態管理に対応せず、ウェブ検索などの組み込みツールも使えないため、store: false を送り、カスタム関数ツールのみを使う必要があります。", "features": [ "サーバーレスで OpenAI 互換の推論：/v1/chat/completions と /v1/responses ルート", "LLM、視覚 LLM、埋め込み、画像生成、音声認識、音声合成を合わせて 40 以上のモデル", "サイズ・価格・能力に応じて最適な利用可能モデルへ振り分ける仮想モデル", "ゼロデータ保持：顧客データをモデルの学習や改善に使わないと OVHcloud は説明", "モデルごとの料金をカタログで 100 万トークンあたりユーロ建てで公開", "透過的なバージョン管理によるモデルのライフサイクル管理。OVHcloud 自社の欧州インフラ上で稼働" ], "use_cases": [ "GPU を管理せずにアプリへチャット・推論・コーディング支援を追加する", "同じ OpenAI 互換キーで埋め込み・画像・音声を取得する", "データ所在地の要件のため推論を欧州インフラ内に保つ" ], "pricing_notes": "サブスクリプションではなく従量課金です。各モデルの 100 万トークンあたり料金が公開カタログに記載され、埋め込みモデルの約 0.01 ユーロから大規模視覚モデルの約 0.91 ユーロまで幅があり、一部のモデル（モデレーションと画像生成）は無料と表示されています。初めて Public Cloud プロジェクトを作成する場合、200 米ドルのトライアルクレジットが提供されます。料金は変動するためカタログが基準であり、本記録は比較可能な paid_from を設定しません。", "faq": [ { "q": "OVHcloud AI Endpoints とは？", "a": "サーバーレス推論 API です。OpenAI 互換の HTTP ルートで、チャット、推論、視覚、埋め込み、画像生成、音声に対応する学習済みオープンウェイトモデルを提供し、アプリ側で GPU を運用する必要をなくします。" }, { "q": "OVHcloud の本社はどこですか？", "a": "フランスのルーベです。OVHcloud の会社情報ページは 1999 年創業の「フランス企業」と述べ、Public Cloud の契約条件には OVH Groupe SA（2 rue Kellermann, 59100 Roubaix）と記載されています。" }, { "q": "料金体系は？", "a": "100 万トークン単位の従量課金で、各モデルの料金はカタログに公開されています。初めての Public Cloud プロジェクトには 200 米ドルのトライアルクレジットが付きます。サブスクリプションは不要です。" }, { "q": "API の制約は？", "a": "OVHcloud のドキュメントによると、/v1/responses ルートは状態管理に対応せず（常に store: false を送信）、ウェブ検索やファイル検索、コード実行などの組み込みツールも利用できません。使えるのはカスタム関数ツールのみです。" }, { "q": "この記録が確立していないことは？", "a": "今回参照したページには、製品自身の対応インターフェース言語、ライセンス条件、地域ごとの提供範囲が記載されていないため、これらの項目は未確認のままです。" } ], "facts": { "ovhcloud-ai-endpoints.hq.france": { "conditions": "ovhcloud.com/en/about-us/ の 2026-10-12 時点の記載による", "note": "ベンダーの「概要」ページで一致: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, "ovhcloud-ai-endpoints.pricing.usage-and-trial": { "value": "100万トークン単位の従量課金でモデルごとの料金を公表。パブリッククラウドの無料トライアルとして US$200 分のクレジット", "conditions": "ovhcloud.com の 2026-10-12 時点の記載による", "note": "製品ページには「US$ 200 in free credit」とあり、カタログはモデルごとの料金を €/Mtoken で記載しています（例：埋め込み 0.01 €、大型ビジュアルモデル 0.91 €）。一部のモデルは「Free」と表示されています。定期プランの価格は公表されていません。" }, "ovhcloud-ai-endpoints.api.openai-routes": { "value": "OpenAI 互換の /v1/chat/completions と /v1/responses ルート", "conditions": "docs.ovhcloud.com の 2026-10-12 時点のドキュメントによる", "note": "ドキュメントは /v1/responses を「the most recent OpenAI-compatible route」と説明し、v1/chat/completions と同様にテキスト生成、マルチターン会話、ツール呼び出しに利用できるとしています。" } }, "selection_evidence": { "tasks": { "note": "レコードから導出：カテゴリ「platforms」。" }, "form_factor": { "note": "ベンダーの製品ページで一致：\"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "note": "レコードの説明・機能・タグに含まれるプラットフォーム語から導出しました。" }, "deployment": { "note": "ベンダーはこのサービスが「on OVHcloud's robust and confidential infrastructure」で動作すると述べており、ネットワーク経由で利用するもので、ユーザーのマシンにはインストールしません。" }, "offline": { "note": "レコードにはオフライン動作の主張が含まれていません。" }, "regions": { "note": "レコードに既にある可用性フィールド（「global」）と一致します。ベンダーは 140 か国で提供していると述べており、リージョン別のサービス一覧は確認していません。" }, "api": { "note": "ベンダーのドキュメントは、/v1/chat/completions や /v1/responses を含む OpenAI 互換の HTTP ルートを記載しています。" }, "license": { "note": "商用製品で、レコードにはソースコードのライセンスが公開されていません。" }, "commercial_use": { "note": "商用製品です。参照したページに非商用の制限は示されていませんが、それは検証されていません。" }, "billing": { "note": "ベンダーはモデル別の従量料率と US$200 のトライアル クレジットを公開していますが、定期的なプラン価格はないため、比較可能な paid_from 値は設定していません。" }, "languages": { "note": "今回のパスで参照したどのベンダー ページにも、製品自身の UI 言語は記載されていません。このキーは、フィールドを空にするのではなく後で回答できるようにするために存在します。" } } }, "de": { "description": "Serverlose Inferenz-API von OVHcloud, die offene Sprach-, Vision-, Embedding-, Bild- und Sprachmodelle über OpenAI-kompatible Endpunkte bereitstellt und nutzungsbasiert abgerechnet wird.", "overview": "OVHcloud AI Endpoints ist ein serverloser Inferenzdienst von OVH Groupe SA, dem französischen Cloud-Anbieter mit Gründung 1999 und Sitz in Roubaix. Entwickler erhalten über eine OpenAI-kompatible API Zugang zu mehr als 40 Modellen mit offenen Gewichten — darunter Llama, Qwen, Mistral, gpt-oss, Whisper und Stable Diffusion — für Chat, Reasoning, Vision, Embeddings, Bildgenerierung, Transkription und Moderation, ohne eigene GPUs zu betreiben. OVHcloud startete den Dienst am 29. April 2025 und rechnet nutzungsbasiert pro Million Tokens ab; neue Public-Cloud-Projekte erhalten ein Testguthaben von 200 US$. Die Einschränkung betrifft die neueste Route: Laut OVHcloud-Dokumentation unterstützt der OpenAI-kompatible Endpunkt /v1/responses weder Zustandsverwaltung noch integrierte Werkzeuge wie Websuche; Aufrufer müssen store: false senden und nur eigene Funktionswerkzeuge nutzen.", "features": [ "Serverlose, OpenAI-kompatible Inferenz: Routen /v1/chat/completions und /v1/responses", "Über 40 Modelle: LLMs, visuelle LLMs, Embeddings, Bildgenerierung, Speech-to-Text und Text-to-Speech", "Virtuelle Modelle, die eine Anfrage nach Größe, Preis oder Fähigkeit dem besten verfügbaren Modell zuordnen", "Keine Datenhaltung: Laut OVHcloud werden Kundendaten nie zum Trainieren oder Verbessern der Modelle verwendet", "Preis pro Modell im Katalog, in Euro pro Million Tokens", "Modell-Lebenszyklus mit transparentem Versioning auf OVHclouds eigener europäischer Infrastruktur" ], "use_cases": [ "Chat, Reasoning oder Coding-Unterstützung in eine Anwendung einbauen, ohne GPUs zu betreiben", "Embeddings, Bilder oder Sprache über denselben OpenAI-kompatiblen Schlüssel beziehen", "Inferenz aus Gründen der Datenresidenz in europäischer Infrastruktur halten" ], "pricing_notes": "Abrechnung nach Nutzung statt Abonnement: Der Preis je Million Tokens steht für jedes Modell im öffentlichen Katalog und reicht von rund 0,01 € für ein Embedding-Modell bis etwa 0,91 € für ein großes visuelles Modell; einige Modelle (Moderation und Bildgenerierung) sind als kostenlos ausgewiesen. OVHcloud gewährt neuen Public-Cloud-Projekten ein Testguthaben von 200 US$. Die Tarife ändern sich, daher gilt der Katalog; dieser Datensatz setzt keinen vergleichbaren paid_from-Wert.", "faq": [ { "q": "Was ist OVHcloud AI Endpoints?", "a": "Eine serverlose Inferenz-API: Sie stellt vortrainierte Modelle mit offenen Gewichten für Chat, Reasoning, Vision, Embeddings, Bildgenerierung und Sprache über OpenAI-kompatible HTTP-Routen bereit, sodass die Anwendung keine eigenen GPUs betreiben muss." }, { "q": "Wo ist OVHcloud ansässig?", "a": "In Roubaix, Frankreich. Die Unternehmensseite nennt OVHcloud ein 1999 gegründetes „französisches Unternehmen“, und die Public-Cloud-Bedingungen nennen OVH Groupe SA, 2 rue Kellermann, 59100 Roubaix." }, { "q": "Wie wird abgerechnet?", "a": "Nutzungsbasiert pro Million Tokens, mit im Katalog veröffentlichten Tarifen je Modell, plus ein Testguthaben von 200 US$ für erste Public-Cloud-Projekte. Ein Abonnement ist nicht erforderlich." }, { "q": "Wo liegen die Grenzen der API?", "a": "Laut OVHcloud-Dokumentation verwaltet die Route /v1/responses keinen Zustand (immer store: false senden) und unterstützt keine integrierten Werkzeuge wie Websuche, Dateisuche oder Codeausführung; nur eigene Funktionswerkzeuge funktionieren." }, { "q": "Was belegt dieser Datensatz nicht?", "a": "Keine der gelesenen Seiten nennt die Oberflächensprachen des Produkts, seine Lizenzbedingungen oder eine Liste der Verfügbarkeit nach Region; diese Felder bleiben daher unbestätigt." } ], "facts": { "ovhcloud-ai-endpoints.hq.france": { "conditions": "Wie auf ovhcloud.com/en/about-us/ am 2026-10-12 angegeben", "note": "Auf der „Über uns“-Seite des Anbieters gefunden: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, "ovhcloud-ai-endpoints.pricing.usage-and-trial": { "value": "Pay-as-you-go pro Million Token mit veröffentlichten Modelltarifen; US$200 Gratis-Testguthaben für die Public Cloud", "conditions": "Wie auf ovhcloud.com am 2026-10-12 angegeben", "note": "Die Produktseite bietet „US$ 200 in free credit\", und der Katalog listet Preise je Modell in €/Mtoken (z. B. Embeddings 0.01 €, ein großes visuelles Modell 0.91 €); einige Modelle sind als „Free\" ausgewiesen. Ein periodischer Tarifpreis wird nicht veröffentlicht." }, "ovhcloud-ai-endpoints.api.openai-routes": { "value": "OpenAI-kompatible Routen /v1/chat/completions und /v1/responses", "conditions": "Wie am 2026-10-12 unter docs.ovhcloud.com dokumentiert", "note": "Die Dokumentation beschreibt /v1/responses als „the most recent OpenAI-compatible route\" und wie v1/chat/completions nutzbar für Textgenerierung, mehrstufige Gespräche und Tool-Aufrufe." } }, "selection_evidence": { "tasks": { "note": "Abgeleitet aus dem Datensatz: Kategorie „platforms“." }, "form_factor": { "note": "Übereinstimmung auf der Produktseite des Anbieters: \"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "note": "Abgeleitet aus Plattformbegriffen in Beschreibung, Funktionen und Tags des Datensatzes." }, "deployment": { "note": "Der Anbieter gibt an, der Dienst laufe „on OVHcloud's robust and confidential infrastructure“; er wird über das Netzwerk erreicht und nicht auf dem Rechner des Nutzers installiert." }, "offline": { "note": "Der Datensatz enthält keine Aussage zum Offline-Betrieb." }, "regions": { "note": "Spiegelt das bereits im Datensatz vorhandene Verfügbarkeitsfeld („global“); der Anbieter gibt an, 140 Länder zu bedienen, und eine regionsbezogene Dienstliste wurde nicht gelesen." }, "api": { "note": "Die Anbieterdokumentation dokumentiert OpenAI-kompatible HTTP-Routen, darunter /v1/chat/completions und /v1/responses." }, "license": { "note": "Kommerzielles Produkt; im Datensatz ist keine Quellcode-Lizenz veröffentlicht." }, "commercial_use": { "note": "Kommerzielles Produkt; die gelesenen Seiten nennen keine nicht-kommerzielle Einschränkung, doch das wurde nicht verifiziert." }, "billing": { "note": "Der Anbieter veröffentlicht modellbezogene Nutzungssätze und ein Testguthaben von US$200, aber keinen periodischen Planpreis, daher ist kein vergleichbarer paid_from-Wert gesetzt." }, "languages": { "note": "Keine in diesem Durchgang gelesene Anbieterseite nennt die eigenen Oberflächensprachen des Produkts; der Schlüssel existiert, damit das Feld später beantwortet werden kann, statt leer zu bleiben." } } }, "pt": { "description": "API de inferência serverless da OVHcloud que serve modelos abertos de linguagem, visão, embeddings, imagem e voz por endpoints compatíveis com a OpenAI, cobrados por uso.", "overview": "OVHcloud AI Endpoints é um serviço de inferência serverless operado pela OVH Groupe SA, a provedora de nuvem francesa fundada em 1999 e sediada em Roubaix. Oferece aos desenvolvedores acesso por API compatível com a OpenAI a mais de 40 modelos de pesos abertos — entre eles Llama, Qwen, Mistral, gpt-oss, Whisper e Stable Diffusion — cobrindo chat, raciocínio, visão, embeddings, geração de imagens, transcrição e moderação, sem implantar GPUs. A OVHcloud lançou o serviço em 29 de abril de 2025 e o cobra por uso por milhão de tokens, com crédito de teste de US$ 200 para novos projetos de Public Cloud. A limitação está na rota mais recente: a documentação da própria OVHcloud diz que o endpoint /v1/responses compatível com a OpenAI não tem estado nem ferramentas integradas como busca na web, sendo preciso enviar store: false e usar apenas funções personalizadas.", "features": [ "Inferência serverless compatível com a OpenAI: rotas /v1/chat/completions e /v1/responses", "Mais de 40 modelos: LLMs, LLMs visuais, embeddings, geração de imagens, fala para texto e texto para fala", "Modelos virtuais que direcionam cada requisição ao melhor modelo disponível por tamanho, preço ou capacidade", "Retenção zero: a OVHcloud afirma que dados de clientes nunca são usados para treinar ou melhorar seus modelos", "Preço por modelo publicado no catálogo, em euros por milhão de tokens", "Gestão do ciclo de vida dos modelos com versionamento transparente, na infraestrutura europeia da OVHcloud" ], "use_cases": [ "Adicionar chat, raciocínio ou assistência de código a um aplicativo sem gerenciar GPUs", "Obter embeddings, imagens ou voz com a mesma chave compatível com a OpenAI", "Manter a inferência em infraestrutura europeia por requisitos de residência de dados" ], "pricing_notes": "Cobrança por uso e não por assinatura: o preço por milhão de tokens de cada modelo é publicado no catálogo, de cerca de € 0,01 para um modelo de embeddings a cerca de € 0,91 para um grande modelo visual, com alguns modelos (moderação e geração de imagens) marcados como gratuitos. A OVHcloud oferece um crédito de teste de US$ 200 para os primeiros projetos de Public Cloud. As tarifas mudam, então o catálogo é a referência; este registro não define nenhum valor paid_from comparável.", "faq": [ { "q": "O que é a OVHcloud AI Endpoints?", "a": "Uma API de inferência serverless: expõe modelos abertos pré-treinados para chat, raciocínio, visão, embeddings, geração de imagens e voz por rotas HTTP compatíveis com a OpenAI, para que o aplicativo não precise operar GPUs próprias." }, { "q": "Onde a OVHcloud está sediada?", "a": "Em Roubaix, na França. Sua página institucional a chama de \"empresa francesa\" fundada em 1999, e os termos de Public Cloud nomeiam a OVH Groupe SA, 2 rue Kellermann, 59100 Roubaix." }, { "q": "Como é cobrado?", "a": "Por uso, por milhão de tokens, com a tarifa de cada modelo publicada no catálogo, mais um crédito de teste de US$ 200 para os primeiros projetos de Public Cloud. Não exige assinatura." }, { "q": "Quais são os limites da API?", "a": "Segundo a documentação da OVHcloud, a rota /v1/responses não gerencia estado (sempre envie store: false) e não oferece ferramentas integradas como busca na web, busca em arquivos ou execução de código; apenas funções personalizadas funcionam." }, { "q": "O que este registro não estabelece?", "a": "Nenhuma página lida informa os idiomas de interface do produto, seus termos de licença nem uma lista de disponibilidade por região, portanto esses campos seguem não confirmados." } ], "facts": { "ovhcloud-ai-endpoints.hq.france": { "conditions": "Conforme indicado em ovhcloud.com/en/about-us/ em 2026-10-12", "note": "Correspondência na página \"Sobre\" do fornecedor: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, "ovhcloud-ai-endpoints.pricing.usage-and-trial": { "value": "Pagamento conforme o uso por milhão de tokens, com tarifas publicadas por modelo; crédito de teste de US$200 na nuvem pública", "conditions": "Conforme indicado em ovhcloud.com em 2026-10-12", "note": "A página do produto oferece \"US$ 200 in free credit\" e o catálogo lista preços por modelo em €/Mtoken (p. ex. embeddings 0.01 €, um modelo visual grande 0.91 €), com alguns modelos apresentados como \"Free\". Não é publicado qualquer preço de plano periódico." }, "ovhcloud-ai-endpoints.api.openai-routes": { "value": "Rotas /v1/chat/completions e /v1/responses compatíveis com OpenAI", "conditions": "Conforme documentado em docs.ovhcloud.com em 2026-10-12", "note": "A documentação descreve /v1/responses como \"the most recent OpenAI-compatible route\" e, tal como v1/chat/completions, utilizável para geração de texto, conversas multi-turno e chamadas de ferramentas." } }, "selection_evidence": { "tasks": { "note": "Derivado do registro: categoria «platforms»." }, "form_factor": { "note": "Correspondência na página do produto do fornecedor: \"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "note": "Derivado das palavras de plataforma na descrição, nos recursos e nas tags do registro." }, "deployment": { "note": "O fornecedor afirma que o serviço roda «on OVHcloud's robust and confidential infrastructure»; ele é acessado pela rede, não instalado na máquina do usuário." }, "offline": { "note": "O registro não contém nenhuma afirmação de funcionamento offline." }, "regions": { "note": "Espelha o campo de disponibilidade já presente no registro («global»); o fornecedor afirma atender 140 países, e nenhuma lista de serviços por região foi lida." }, "api": { "note": "A documentação do fornecedor documenta rotas HTTP compatíveis com OpenAI, incluindo /v1/chat/completions e /v1/responses." }, "license": { "note": "Produto comercial; nenhuma licença do código-fonte é publicada no registro." }, "commercial_use": { "note": "Produto comercial; as páginas consultadas não indicam restrição não comercial, mas isso não foi verificado." }, "billing": { "note": "O fornecedor publica tarifas de uso por modelo e um crédito de teste de US$200, mas nenhum preço de plano periódico, então nenhum valor paid_from comparável é definido." }, "languages": { "note": "Nenhuma página do fornecedor lida nesta passagem indica os idiomas de interface do próprio produto; a chave existe para que o campo possa ser respondido mais tarde em vez de ficar ausente." } } }, "ko": { "description": "OVHcloud의 서버리스 추론 API로, OpenAI 호환 엔드포인트를 통해 오픈 웨이트 언어·비전·임베딩·이미지·음성 모델을 제공하고 사용량 기준으로 과금한다.", "overview": "OVHcloud AI Endpoints는 1999년 설립되어 프랑스 루베에 본사를 둔 클라우드 사업자 OVH Groupe SA가 운영하는 서버리스 추론 서비스입니다. 개발자는 OpenAI 호환 API로 Llama, Qwen, Mistral, gpt-oss, Whisper, Stable Diffusion 등 40개 이상의 오픈 웨이트 모델에 접근해 GPU를 직접 운영하지 않고도 채팅, 추론, 비전, 임베딩, 이미지 생성, 전사, 모더레이션을 구현할 수 있습니다. OVHcloud는 2025년 4월 29일에 이 서비스를 출시했고 100만 토큰당 종량제로 과금하며, 신규 Public Cloud 프로젝트에는 미화 200달러 체험 크레딧을 제공합니다. 제약은 최신 라우트에 있습니다. OVHcloud 문서에 따르면 OpenAI 호환 /v1/responses 엔드포인트는 상태 관리를 지원하지 않고 웹 검색 같은 내장 도구도 제공하지 않으므로 store: false를 보내고 사용자 정의 함수 도구만 사용해야 합니다.", "features": [ "서버리스 OpenAI 호환 추론: /v1/chat/completions와 /v1/responses 라우트", "LLM, 비전 LLM, 임베딩, 이미지 생성, 음성 인식, 음성 합성을 아우르는 40개 이상의 모델", "크기·가격·성능 기준으로 요청을 가장 적합한 사용 가능 모델에 매핑하는 가상 모델", "제로 데이터 보존: 고객 데이터를 모델 학습이나 개선에 사용하지 않는다고 OVHcloud는 밝힘", "모델별 가격을 카탈로그에 100만 토큰당 유로로 공개", "투명한 버전 관리로 모델 수명주기를 관리하며 OVHcloud 자체 유럽 인프라에서 운영" ], "use_cases": [ "GPU를 관리하지 않고 애플리케이션에 채팅·추론·코딩 지원을 추가", "동일한 OpenAI 호환 키로 임베딩·이미지·음성을 얻기", "데이터 거주 요건을 위해 추론을 유럽 인프라 안에 유지" ], "pricing_notes": "구독이 아니라 사용량 기준으로 과금합니다. 모델별 100만 토큰당 가격이 공개 카탈로그에 있으며, 임베딩 모델 약 0.01유로부터 대형 비전 모델 약 0.91유로까지이고 일부 모델(모더레이션과 이미지 생성)은 무료로 표시됩니다. 처음 Public Cloud 프로젝트를 만드는 경우 미화 200달러 체험 크레딧이 제공됩니다. 요금은 변동되므로 카탈로그가 기준이며, 이 기록은 비교 가능한 paid_from 값을 설정하지 않습니다.", "faq": [ { "q": "OVHcloud AI Endpoints는 무엇인가요?", "a": "서버리스 추론 API입니다. OpenAI 호환 HTTP 라우트로 채팅, 추론, 비전, 임베딩, 이미지 생성, 음성을 위한 사전 학습된 오픈 웨이트 모델을 제공해 애플리케이션이 자체 GPU를 운영할 필요를 없앱니다." }, { "q": "OVHcloud는 어디에 있나요?", "a": "프랑스 루베입니다. 회사 소개 페이지는 1999년 설립된 \"프랑스 기업\"이라고 밝히고, Public Cloud 약관에는 OVH Groupe SA(2 rue Kellermann, 59100 Roubaix)가 명시되어 있습니다." }, { "q": "어떻게 과금되나요?", "a": "100만 토큰당 종량제이며 각 모델 요금은 카탈로그에 공개됩니다. 첫 Public Cloud 프로젝트에는 미화 200달러 체험 크레딧이 제공됩니다. 구독은 필요하지 않습니다." }, { "q": "API의 제약은 무엇인가요?", "a": "OVHcloud 문서에 따르면 /v1/responses 라우트는 상태를 관리하지 않고(항상 store: false 전송) 웹 검색, 파일 검색, 코드 실행 같은 내장 도구도 지원하지 않습니다. 사용자 정의 함수 도구만 쓸 수 있습니다." }, { "q": "이 기록이 확립하지 않는 것은?", "a": "이번에 읽은 페이지에는 제품 자체의 인터페이스 언어, 라이선스 조건, 지역별 제공 범위가 적혀 있지 않아 해당 항목은 확인되지 않은 상태입니다." } ], "facts": { "ovhcloud-ai-endpoints.hq.france": { "conditions": "ovhcloud.com/en/about-us/의 2026-10-12 기준 기재에 따름", "note": "공급업체 \"소개\" 페이지에서 일치: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, "ovhcloud-ai-endpoints.pricing.usage-and-trial": { "value": "백만 토큰당 종량제이며 모델별 요율을 공개합니다. 퍼블릭 클라우드 무료 체험 크레딧 US$200", "conditions": "ovhcloud.com의 2026-10-12 기준 기재에 따름", "note": "제품 페이지는 \"US$ 200 in free credit\"을 제공하며, 카탈로그는 모델별 가격을 €/Mtoken으로 표시합니다(예: 임베딩 0.01 €, 대형 비주얼 모델 0.91 €). 일부 모델은 \"Free\"로 표시됩니다. 정기 요금제 가격은 공개되지 않았습니다." }, "ovhcloud-ai-endpoints.api.openai-routes": { "value": "OpenAI 호환 /v1/chat/completions 및 /v1/responses 라우트", "conditions": "docs.ovhcloud.com의 2026-10-12 기준 문서에 따름", "note": "문서는 /v1/responses를 \"the most recent OpenAI-compatible route\"라고 설명하며, v1/chat/completions와 마찬가지로 텍스트 생성, 다중 턴 대화, 도구 호출에 사용할 수 있다고 합니다." } }, "selection_evidence": { "tasks": { "note": "레코드에서 도출: 카테고리 “platforms”." }, "form_factor": { "note": "벤더의 제품 페이지에서 일치: \"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "note": "레코드의 설명·기능·태그에 있는 플랫폼 관련 단어에서 도출했습니다." }, "deployment": { "note": "벤더는 이 서비스가 “on OVHcloud's robust and confidential infrastructure”에서 실행된다고 밝힙니다. 이는 네트워크를 통해 접근하는 것이며 사용자 기기에 설치하는 것이 아닙니다." }, "offline": { "note": "레코드에는 오프라인 작동에 관한 주장이 없습니다." }, "regions": { "note": "레코드에 이미 있는 가용성 필드(“global”)와 일치합니다. 벤더는 140개국에 서비스를 제공한다고 밝혔으며, 지역별 서비스 목록은 확인하지 않았습니다." }, "api": { "note": "벤더 문서는 /v1/chat/completions 및 /v1/responses를 포함한 OpenAI 호환 HTTP 경로를 문서화하고 있습니다." }, "license": { "note": "상업용 제품이며, 레코드에 소스 라이선스가 공개되어 있지 않습니다." }, "commercial_use": { "note": "상업용 제품입니다. 확인한 페이지에 비상업적 제한은 명시되어 있지 않지만, 이는 검증되지 않았습니다." }, "billing": { "note": "벤더는 모델별 사용 요율과 US$200 체험 크레딧을 공개하지만 정기 요금제 가격은 없어, 비교 가능한 paid_from 수치를 설정하지 않았습니다." }, "languages": { "note": "이번 확인에서 읽은 어떤 벤더 페이지에도 제품 자체의 인터페이스 언어가 명시되어 있지 않습니다. 이 키는 필드를 비워 두지 않고 나중에 답할 수 있도록 존재합니다." } } } }, "sources": [ { "title": "OVHcloud AI Endpoints", "url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/" }, { "title": "About us", "url": "https://www.ovhcloud.com/en/about-us/" }, { "title": "AI Endpoints model catalogue and prices", "url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/catalog" }, { "title": "Public Cloud price list", "url": "https://www.ovhcloud.com/en/public-cloud/prices/" }, { "title": "AI Endpoints — Responses API (documentation)", "url": "https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-responses-api" }, { "title": "OVHcloud launches AI Endpoints (press release, 29 April 2025)", "url": "https://corporate.ovhcloud.com/en/newsroom/news/ai-endpoints" }, { "title": "OVHcloud — Wikipedia", "url": "https://en.wikipedia.org/wiki/OVHcloud" } ], "facts": [ { "id": "ovhcloud-ai-endpoints.hq.france", "field": "hq_country", "value": "FR", "conditions": "As stated on ovhcloud.com/en/about-us/ on 2026-10-12", "sources": [ "https://www.ovhcloud.com/en/about-us/" ], "checked": "2026-10-12", "evidence_type": "official_page", "status": "verified", "note": "Matched on the vendor about page: \"Founded in 1999, OVHcloud is a French company with an international presence\". The Public Cloud terms on the AI Endpoints page name \"OVH Groupe SA (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France)\"." }, { "id": "ovhcloud-ai-endpoints.pricing.usage-and-trial", "field": "pricing", "value": "Pay-as-you-go per million tokens, per-model rates published; US$200 Public Cloud free-trial credit", "conditions": "As stated on ovhcloud.com on 2026-10-12", "sources": [ "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/", "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/catalog" ], "checked": "2026-10-12", "evidence_type": "official_page", "status": "verified", "note": "The product page offers \"US$ 200 in free credit\" and the catalogue lists per-model prices in €/Mtoken (e.g. embeddings 0.01 €, a large visual model 0.91 €), with a few models shown as \"Free\". No periodic plan price is published." }, { "id": "ovhcloud-ai-endpoints.api.openai-routes", "field": "api", "value": "OpenAI-compatible /v1/chat/completions and /v1/responses routes", "conditions": "As documented at docs.ovhcloud.com on 2026-10-12", "sources": [ "https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-responses-api" ], "checked": "2026-10-12", "evidence_type": "official_page", "status": "verified", "note": "The documentation describes /v1/responses as \"the most recent OpenAI-compatible route\" and, like v1/chat/completions, usable for text generation, multi-turn conversations and tool calling." } ], "updated": "2026-10-12", "selection": { "tasks": [ "model-hosting" ], "form_factor": "api", "platforms": [ "web" ], "deployment": [ "cloud" ], "offline": "unknown", "regions": [ "global" ], "api": "yes", "license": "proprietary", "commercial_use": "unknown", "billing": { "models": [ "usage" ], "free_allowance": "limited", "requires_subscription": "no" }, "languages": [ "unknown" ] }, "selection_evidence": { "tasks": { "status": "editorial_inference", "note": "Derived from the record: category \"platforms\"." }, "form_factor": { "status": "verified", "source": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/", "checked": "2026-10-12", "basis": "official_page", "note": "Matched on the vendor product page: \"AI Endpoints is the serverless inference API for a selection of models\"." }, "platforms": { "status": "editorial_inference", "note": "Derived from platform words in the record description/features/tags." }, "deployment": { "status": "verified", "source": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/", "checked": "2026-10-12", "basis": "official_page", "note": "The vendor states the service runs \"on OVHcloud's robust and confidential infrastructure\"; it is reached over the network, not installed on the user's machine." }, "offline": { "status": "editorial_inference", "note": "The record contains no offline-operation claim." }, "regions": { "status": "editorial_inference", "note": "Mirrors the availability field already on the record (\"global\"); the vendor says it serves 140 countries, and no per-region service list was read." }, "api": { "status": "verified", "source": "https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-responses-api", "checked": "2026-10-12", "basis": "official_page", "note": "The vendor documentation documents OpenAI-compatible HTTP routes, including /v1/chat/completions and /v1/responses." }, "license": { "status": "editorial_inference", "note": "Commercial product; no source licence is published in the record." }, "commercial_use": { "status": "editorial_inference", "note": "Commercial product; the pages read do not state a non-commercial restriction, but that was not verified." }, "billing": { "status": "editorial_inference", "note": "The vendor publishes per-model usage rates and a US$200 trial credit, but no periodic plan price, so no comparable paid_from figure is set." }, "languages": { "status": "unknown", "note": "No vendor page read in this pass states the product's own interface languages; the key exists so the field can be answered later rather than left absent." } }, "verification": { "last_checked": "2026-10-12", "batch": "cp18", "channel": "Page text was read this session with the available page-extraction tool; the design-canvas browser returned session_not_registered on every attempt (see docs/coverage-memoranda/non-us-coverage.md §8.1).", "pages": [ { "url": "https://www.ovhcloud.com/en/about-us/", "status": 200, "fetched_at": "2026-10-12T04:41:45.000Z" }, { "url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/", "status": 200, "fetched_at": "2026-10-12T04:41:45.000Z" }, { "url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/catalog", "status": 200, "fetched_at": "2026-10-12T04:41:45.000Z" }, { "url": "https://www.ovhcloud.com/en/public-cloud/prices/", "status": 200, "fetched_at": "2026-10-12T04:41:45.000Z" }, { "url": "https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-responses-api", "status": 200, "fetched_at": "2026-10-12T04:41:45.000Z" }, { "url": "https://corporate.ovhcloud.com/en/newsroom/news/ai-endpoints", "status": 200, "fetched_at": "2026-10-12T04:41:45.000Z" }, { "url": "https://en.wikipedia.org/wiki/OVHcloud", "status": 200, "fetched_at": "2026-10-12T04:41:45.000Z" } ], "unreachable": [ "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/pricing/ (404 page not found)", "https://help.ovhcloud.com/csm/en-public-cloud-ai-endpoints-getting-started (fetch error)" ], "confirmed_fields": [ "hq_country", "api", "billing", "deployment" ], "gaps": [ "license", "regions", "languages", "commercial_use" ], "unconfirmed_fields": [ "tasks", "platforms", "offline", "regions", "license", "commercial_use" ], "gap_note": "The pages fetched on this date carry no statement of the product's own interface languages, its licence terms, or a per-region service list; those fields keep their inferred or unknown status rather than being presented as confirmed. The separate /pricing/ path returned 404, so the catalogue URL is recorded as the pricing page instead of inventing one." }, "url": "https://globalaiproductindex.com/products/ovhcloud-ai-endpoints/" }