上游供应网络
用于支撑我们统一 AI API 服务的内部采购与路由视图
Amazon 的全托管服务,通过单一 API 提供 Anthropic、Meta、Mistral、Cohere 和 Amazon 自有 Titan 模型的广泛基础模型访问,支持企业安全和自定义选项。
Cloud platform accessProcurement readyRecommended未实时验证
运营模式
Cloud platform access
接入结论
Best when your team already runs on the matching cloud and wants unified billing, IAM, and support.
unknown · 14d cadence
Groq delivers ultra-fast LLM inference using custom Language Processing Units (LPUs), supporting open models like Llama and Mixtral with exceptional speed for real-time applications.
Cloud platform accessProcurement readyRecommended未实时验证
运营模式
Cloud platform access
接入结论
Best when your team already runs on the matching cloud and wants unified billing, IAM, and support.
unknown · 14d cadence
SiliconFlow 提供高性能的一体化 AI 云平台,具有统一 API,用于开源多模态模型的快速推理,强调速度和成本效益。
Cloud platform accessProcurement readyRecommended未实时验证
运营模式
Cloud platform access
接入结论
Best when your team already runs on the matching cloud and wants unified billing, IAM, and support.
unknown · 14d cadence
快速启动
$npx ccjk -p siliconflow-api
Together AI是一个云平台,旨在运行和微调开源AI模型,具有快速推理和有竞争力的定价。该平台提供对50多个流行开源模型的访问,包括Llama 2、Mistral、Mixtral、CodeLlama和Yi,支持自定义模型微调和部署。Together AI为新用户提供5美元的免费额度来探索平台,以及开发友好的API、全面的文档和企业级基础设施,适用于需要灵活开源AI部署的研究和生产环境。
Cloud platform accessProcurement readyRecommended未实时验证
运营模式
Cloud platform access
接入结论
Best when your team already runs on the matching cloud and wants unified billing, IAM, and support.
unknown · 14d cadence
支持模型:
llama-2mistralmixtral+2
Groq是一个超快AI推理平台,利用定制设计的LPU(语言处理单元)硬件为开源LLM提供前所未有的推理速度。该平台通过OpenAI兼容的API提供对Llama 2、Mixtral和Gemma等流行模型的免费访问,使开发人员能够轻松地将超快AI能力集成到他们的应用程序中。Groq的定制硬件实现了比传统GPU快10倍的令牌生成速度,具有慷慨的免费层级和有竞争力的按使用付费定价,适用于需要最大性能的生产工作负载。
Cloud platform accessProcurement guardedRecommended校验受阻
运营模式
Cloud platform access
接入结论
Good production candidate when low-latency managed inference on GroqCloud matters more than direct control over every open model host. Live verification is currently partial because some required official source types are blocked from this environment: documentation, pricing, support, terms.
overdue · 21d cadence
实时校验
Required official source types exist, but live verification is currently blocked from this environment or region.
0 verified · 4 blocked · 0 broken
official baseline
支持模型:
llama-3.3-70bmixtral-8x7bgemma-7b
AWS托管的基础模型服务,支持多个提供商
Cloud platform accessProcurement readyRecommended未实时验证
运营模式
Cloud platform access
接入结论
Best when your team already runs on the matching cloud and wants unified billing, IAM, and support.
unknown · 14d cadence
支持模型:
claude-3llama-2titan+1
快速启动
$npx ccjk -p amazon-bedrock
腾讯的混元模型,与微信和腾讯云深度集成
Cloud platform accessProcurement readyRecommended未实时验证
运营模式
Cloud platform access
接入结论
Best when your team already runs on the matching cloud and wants unified billing, IAM, and support.
unknown · 14d cadence
支持模型:
hunyuan-litehunyuan-standardhunyuan-pro
快速启动
$npx ccjk -p tencent-hunyuan