AMD Radeon Cloud (Token Factory)

Verified ✅ Truly Free

Token Factory on AMD's Radeon Cloud China site is a beta service that hosts a rotating set of models on Radeon GPUs and serves them through an OpenAI-compatible API (it also accepts the Anthropic Messages format). Usage is free but capped: 20 requests/minute per account, 30 per API key, 8 concurrent requests per key, 120 per IP, plus a daily allowance counted in points that resets at 00:00 Beijing time. AMD describes it as a free trial and development service, not for production, and the model catalog changes over time.

OpenAI-Compatible No credit card Free Quota
Get API Key Suggest Edit
0

Overview

Provider Type

API

API Endpoint

https://developer.amd.com.cn/radeon/api/v1

Free Tier Highlights

20 RPM per account30 RPM per API key120 RPM per IP8 concurrent requests per key

Why Choose AMD Radeon Cloud (Token Factory)?

Free, key-only access to several strong open models (GLM, Qwen, DeepSeek, MiniCPM) through an OpenAI- and Anthropic-compatible endpoint, with a daily allowance that resets every day.

Quick Start Guide

1

Create an AMD AI Developer Program account at https://developer.amd.com.cn and sign in to Radeon Cloud

2

Open Token Factory in the console and copy your API key and base URL

3

Call https://developer.amd.com.cn/radeon/api/v1 with an OpenAI-compatible client, or GET /v1/models to list current models

Available Models

7 free models
Model Name ID Context Capabilities
GLM-5.3-Flash Free -
See provider
-
MiMo-V2.6-Flash Free -
See provider
-
Qwen3.8-Flash-Next Free -
See provider
-
DeepSeek-V4.1-Flash Free -
See provider
-
MiniCPM-V-4.7-35B-A3B Free -
See provider
Vision
Gemma-4-31B-IT Free -
See provider
-
Qwen3.8-27B Free -
See provider
-

Free Tier Pricing & Limits

Rate Limit

Requests per minute

20 Requests per minute per account30 Requests per minute per API key120 Requests per minute per IP8 concurrent requests per key

Daily Quota

Requests per day

Daily cap in points (pts)per account; resets 00:00 UTC+8

Token Limit

Tokens per minute

See provider

Monthly Quota

Per month limit

None specified

Use Cases

Prototyping and development

Coding assistants (Claude Code and similar clients)

Testing multiple open models with one key

Limitations & Considerations

China site only; a separate global site exists without Token Factory

Model catalog changes over time - resolve models at runtime

Free trial/development service, not for production use; reselling not allowed

Some new accounts are reviewed before an API key is issued

Exact daily point allowance is per account and shown in the console

Community Hub

Live

Join the discussion, share tips, and rate AMD Radeon Cloud (Token Factory).

Quick Reactions

Add Discussion

Comments are moderated. Be helpful and respectful.

Recent Activity

0 comments

Ready to Get Started?

Join thousands of developers using AMD Radeon Cloud (Token Factory)

Start Building Now