AMD Radeon Cloud Public Free Model APIs
Operated by AMD developer community / Radeon Cloud
No-cost research and development inference across 9 models; not unrestricted production use.
https://developer.amd.com.cn/radeon/api/v1Models mentioned
DeepSeek-V4-FlashDeepSeek-V4-Flash-Vision-ExpDeepSeek-V4.1-FlashQwen3.8-Flash-NextQwen3.8-27BGLM-5.3-FlashMiMo-V2.6-FlashMiniCPM5-2BMinerU2.5-ProEquivalent paid value
How this was valued: Modeled admission-only throughput upper bound for the exact served DeepSeek-V4-Flash-0731 checkpoint. The tighter account gateway limit is 20 RPM, and the official endpoint documents 1048576 total input-plus-output tokens per request. The current exact-model OpenRouter comparison is $0.0103/M input and $1.28/M output; the comparison's 943718-output-token ceiling is modeled for the hosted endpoint, leaving 104858 input tokens per full request. Assumes uninterrupted successful saturation with no latency, eight-request-concurrency, shared-capacity, or fair-use loss. This is not the actual account grant, expected delivered value, or a cash balance; the hidden rolling 24-hour USD budget can reduce it sharply.
Partial because the authenticated-only per-account spend ceiling is unpublished and determines actual deliverable usage; the separate endpoint output cap is modeled, and routes without an exact paid comparison are excluded. The public ceiling remains an upper bound, not proof that the account can consume it.
Limits and terms
- Platform Requests Per Minute Per Key
- 30
- Platform Requests Per Minute Per Ip
- 120
- Platform Concurrency Per Key
- 8
- Gateway Requests Per Minute Per Account
- 20
- Gateway Window Seconds
- 60
- Spend Cap
- dynamic account specific usd amount not public
- Effective Note
- The 20 RPM account gateway is tighter than the typical 30 RPM key admission limit.
- Spend Window
- Rolling 24 hours starting with the first billed request, not a calendar-day reset; account-specific budget remains unpublished.
What happens to your prompts?
AMD ordinarily retains prompts and model outputs for a rolling two days, despite excluding training, sale, and third-party own use.
- Plan Scope
- AMD Radeon Cloud public free Model API for development, evaluation, demos, hackathons, and open-source work.
- Prompt Retention
- AMD retains prompts in ordinary operation for a rolling two-day period; abuse investigations or legal obligations can extend retention.
- Response Retention
- Model outputs follow the same rolling two-day content-retention rule and possible abuse/legal holds.
- Ordinary Logging
- Request metadata, account/key, model, token, timing, rate-limit, security, and operational records may be retained longer than content; no universal metadata maximum is stated.
- Model Training
- AMD says prompts and outputs are never used to train models.
- Product Improvement
- Content may be used during the two-day window for capacity planning, debugging, abuse prevention, and enforcement, but not sold or licensed for third-party own use.
- Human Or Operator Access
- Authorized personnel may access short-lived content for debugging, abuse investigation, security, and legal compliance.
- Subprocessors And Routing
- AMD/Radeon Cloud operates the free gateway and deployments; the terms do not grant unrelated third parties independent content use.
- Deletion Controls
- Ordinary prompt/output copies roll off after two days; users have no documented per-request deletion control, and abuse/legal holds can extend retention.
- Caveat
- No sensitive or confidential data may be submitted. Short retention and no training are favorable, but default ordinary content logging requires a not private classification under this catalog's rubric.
Governing documents
- Model API Terms
- https://amd-aim.github.io/radeon-cloud-docs/api/terms/
Eligibility
- Account Required
- Yes
- Signup
- Register or sign in to the AMD Developer/Token Factory account; the API key is issued automatically on first registration.
- Signup Open To Anyone
- Yes
- Keys Per Person
- 1
- Payment Method Required
- No
- Real Name Verification Required
- not documented
Primary sources
Before you build with AMD Radeon Cloud Public Free Model APIs
Read the classification narrowly
This record has current first-party evidence for some zero-cost hosted inference. The label describes the bounded offer supported by evidence on 2026-10-06; it does not guarantee permanence, production suitability, uptime, latency, model quality, or access from every region.
Resolve the live model route
This snapshot records 9 model IDs. Match the exact ID against the provider's live catalog before using it in code; zero-price routes and aliases can rotate while an older documentation page remains online. The recorded base endpoint is https://developer.amd.com.cn/radeon/api/v1, but the provider's current API reference remains authoritative for paths, authentication, and request shape.
Confirm account and billing boundaries
An account is required, so public catalog evidence cannot by itself prove the limits, balance, or billing behavior attached to your workspace. The reviewed evidence explicitly says no payment card is required. Re-check the signup flow because eligibility and billing controls can change after the snapshot. The equivalent paid value shown here prices a documented allowance where possible; it is not cash, guaranteed savings, or protection from overage.
Re-read the prompt policy for your plan
This record classifies the reviewed handling as “Not private”: AMD ordinarily retains prompts and model outputs for a rolling two days, despite excluding training, sale, and third-party own use. Provider policies can distinguish free, paid, enterprise, opted-in, and feature-specific traffic, so verify the governing terms for the exact account and route that will receive your data.
Follow the evidence, then re-check it
The record links 9 first-party sources covering the offer, catalog, limits, pricing, terms, privacy, or adoption evidence available to the audit. Prefer the newest governing document or live catalog when sources disagree, and submit a correction when a provider changes a material term.
Match the quota to the workload shape
Translate the published allowance into the traffic pattern you actually expect instead of comparing headline totals alone. A daily token pool can look generous while a low requests-per-minute or concurrency ceiling blocks interactive bursts; a high request limit can still fail a long-context job when input, output, or per-request tokens are capped. Separate prompt tokens from generated tokens, include retries and tool calls, and test the largest realistic payload. If the provider documents more than one limit window, the tightest window at your peak load is the practical ceiling. Research, experimental, and community access can also carry eligibility or fair-use constraints that cannot be modeled as a simple number.
Plan fallback without changing the rules
A fallback should preserve more than API syntax. Confirm that the substitute route supports the required modality, context length, streaming behavior, structured output, tools, and safety controls, then compare its prompt-retention and training terms. An OpenAI-compatible request shape does not make providers operationally or contractually equivalent. Decide which errors may trigger a retry, cap retry storms, and prevent an exhausted free route from silently switching to a billable model. If deterministic output matters, record the model revision and sampling settings; rotating aliases and free-model pools can change behavior even when the endpoint remains available.
Monitor the offer as a dependency
Capture the model ID, response model field, rate-limit headers, usage fields, latency, HTTP status, and any provider request identifier for each test. Watch for authorization failures, quota exhaustion, catalog removal, policy revisions, and dashboard balance changes as separate failure modes. Re-check the provider’s live catalog and governing pages on a schedule proportionate to the workload’s importance, and keep the dated sources that supported your decision. Free capacity is especially suitable for prototypes, evaluations, fallbacks, and bounded workloads when the application can tolerate change; a production dependency still needs observability, an exit path, and an owner responsible for re-verification.
Test one complete request before scaling
Start with the smallest permitted request using the exact credential, model ID, endpoint, and account type you intend to deploy. Record the HTTP status, response headers, usage fields, latency, and any dashboard balance change. Then exercise the failure path: an invalid model, an exhausted quota, or a rate-limit response should fail clearly without silently switching to a paid route. If the provider supports streaming or tool calls, validate those features separately because a free model can expose a narrower capability set than its paid counterpart. Keep a budget ceiling outside the application whenever billing is possible, and avoid sending sensitive data until the observed route matches the reviewed data agreement.