这个资源提供什么What this offer provides
AMD Radeon Cloud 官方文档将 Public Free Model APIs 定义为常驻共享端点：无需创建 Dedicated 实例、不消耗 credits；一个 Token Factory API Key 可访问当前共享模型目录。官方模型概览当前列出 6 个聊天模型，最高 1,048,576 tokens 上下文。AMD Radeon Cloud Public Free Model APIs ： Dedicated 、 credits； Token Factory API Key 。 6 ， 1,048,576 tokens 。
关键信息Key details
免费方式Free method6 个当前共享聊天模型 · 无需创建实例 · 不消耗 credits · 最高 1,048,576 tokens 上下文6 current shared chat models · no instance · no credits · up to 1,048,576-token context
有效期ValidityAlways-on shared tier; AMD may add or retire models at any timeAlways-on shared tier; AMD may add or retire models at any time
访问条件AccessRadeon Cloud account verification + rc- API key; shared rate limits applyRadeon Cloud account verification + rc- API key; shared rate limits apply
最后核验Last checked2026-09-22
快速上手Quick start
登录 Radeon Cloud Token Factory 获取 rc- API Key，调用 AMD 的共享免费模型 APIFollow the model-specific steps to register, create a key and connect the model.
✓
注册与接入步骤Registration and access steps
准备条件Prerequisites
- AMD Radeon Cloud account
- Account verification completed
- rc- API key from Token Factory
- 打开 Token Factory 并登录 Radeon Cloud 账号，完成平台账户验证。Token Factory Radeon Cloud ， 。
- 在 Token Factory 获取 rc- 前缀 API Key；同一个 Key 可访问当前共享免费模型。Token Factory rc- API Key； Key 。
- 先请求 GET https://developer.amd.com.cn/radeon/api/v1/models，确认当前 free=true 的模型与实际 model ID。GET https://developer.amd.com.cn/radeon/api/v1/models， free=true model ID。
- 按 OpenAI 兼容的 /chat/completions 或 Anthropic 兼容的 /messages 发送请求，并遵守共享限流和滚动 spend cap。OpenAI /chat/completions Anthropic /messages ， spend cap。
4
发送第一条请求Send first request
使用下方命令开始调用。Use the command below to make your first request.
可复制命令Copyable command
export RADEON_API_KEY="你的 Key"
curl https://developer.amd.com.cn/radeon/api/v1/chat/completions --header "Authorization: Bearer $RADEON_API_KEY" --header 'Content-Type: application/json' --data '{"model":"DeepSeek-V4-Flash","messages":[{"role":"user","content":"你好"}]}'免费模型逐个看Free models by entry
| 模型Model | 上下文窗口Context window | 免费额度Free quota | 备注Notes |
|---|---|---|---|
DeepSeek-V4-FlashDeepSeek-V4-FlashText · reasoningText · reasoning | 1,048,576 tokens1,048,576 tokens | Typical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend capTypical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend cap | API model ID is DeepSeek-V4-Flash; AMD documents the served weights as DeepSeek-V4-Flash-0731.API model ID is DeepSeek-V4-Flash; AMD documents the served weights as DeepSeek-V4-Flash-0731. |
DeepSeek-V4-Flash-Vision-ExpDeepSeek-V4-Flash-Vision-ExpVision · experimentalVision · experimental | 1,048,576 tokens1,048,576 tokens | Typical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend capTypical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend cap | Multimodal shared endpoint for text and image input.Multimodal shared endpoint for text and image input. |
DeepSeek-V4.1-FlashDeepSeek-V4.1-FlashVision · reasoningVision · reasoning | 1,048,576 tokens1,048,576 tokens | Typical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend capTypical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend cap | Current AMD shared public model listed in the official model overview.Current AMD shared public model listed in the official model overview. |
Qwen3.8-Flash-NextQwen3.8-Flash-NextVision · reasoning · experimentalVision · reasoning · experimental | 262,144 tokens262,144 tokens | Typical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend capTypical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend cap | Experimental multimodal Qwen endpoint in the current public-free catalog.Experimental multimodal Qwen endpoint in the current public-free catalog. |
Qwen3.8-27BQwen3.8-27BVision · reasoningVision · reasoning | 131,072 tokens131,072 tokens | Typical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend capTypical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend cap | Current Qwen 3.8 27B public-free endpoint.Current Qwen 3.8 27B public-free endpoint. |
MiniCPM5-2BMiniCPM5-2BText · lightweightText · lightweight | 131,072 tokens131,072 tokens | Typical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend capTypical public-free limits: 30 RPM/key · 120 RPM/IP · 8 concurrent/key; gateway 20 RPM/account plus rolling spend cap | MiniCPM5-1B is no longer published; AMD directs this size class to MiniCPM5-2B.MiniCPM5-1B is no longer published; AMD directs this size class to MiniCPM5-2B. |
额度以官方页面和控制台实时显示为准；公测或限免额度可能随时调整。Quota follows the official page and console; preview or limited free access can change.
如何使用How to use
登录 Radeon Cloud Token Factory 获取 rc- API Key，调用 AMD 的共享免费模型 APIRadeon Cloud Token Factory rc- API Key， AMD API
前置条件Prerequisites
- AMD Radeon Cloud accountAMD Radeon Cloud account
- Account verification completedAccount verification completed
- rc- API key from Token Factoryrc- API key from Token Factory
步骤Steps
- 打开 Token Factory 并登录 Radeon Cloud 账号，完成平台账户验证。Token Factory Radeon Cloud ， 。
- 在 Token Factory 获取 rc- 前缀 API Key；同一个 Key 可访问当前共享免费模型。Token Factory rc- API Key； Key 。
- 先请求 GET https://developer.amd.com.cn/radeon/api/v1/models，确认当前 free=true 的模型与实际 model ID。GET https://developer.amd.com.cn/radeon/api/v1/models， free=true model ID。
- 按 OpenAI 兼容的 /chat/completions 或 Anthropic 兼容的 /messages 发送请求，并遵守共享限流和滚动 spend cap。OpenAI /chat/completions Anthropic /messages ， spend cap。
FreeLLM TEST
预检通过 · 登录态实测待补Preflight passed · signed-in test pendingFreeLLM 自测FreeLLM test
测试时间Tested at2026-09-22
测试级别Test levelpreflight
真实登录使用Signed-in usage未验证Not verified
我们测了什么What we tested
核对 AMD Token Factory 免费共享端点、当前模型目录、鉴权方式、双协议兼容与限流规则。
- 读取 AMD Radeon Cloud 官方 API 概览、模型目录、模型参考、鉴权和限流文档。
- 交叉核对 Token Factory 官方入口与官方请求示例，确认 Base URL 和模型 ID。
测试结果Result
官方资料确认 Public Free Model APIs 常驻、零 credits，并列出 6 个当前聊天模型；本次未使用真实 Radeon Cloud API Key 发起模型生成请求。
未验证 / 限制：Not verified / limits 当前会话没有可用的 Radeon Cloud 登录态或 rc- API Key，因此未验证账户内 Key 展示、实际生成响应和该账户实时 spend-cap 数值。
测试证据Test evidence
官方来源Official sources
AMD Radeon Cloud 官方文档确认 Public Free Model APIs 为 always-on 共享端点，免费使用且不消耗 credits。官方模型概览（2026-09-15 更新）当前列出 6 个聊天模型，并明确 MiniCPM5-1B 已停止发布。官方限流文档给出典型值：每 Key 30 RPM、每 IP 120 RPM、每 Key 并发 8，并另有账户级网关限流和滚动 spend cap。AMD Radeon Cloud Public Free Model APIs always-on ， credits。 （2026-09-15 ） 6 ， MiniCPM5-1B 。 ： Key 30 RPM、 IP 120 RPM、 Key 8， spend cap。