FreeTheAI migration: fix .xyz URL and model 404

If an older FreeTheAI setup stopped working, replace the retired API host and model ID. FreeTheAI's current docs say old api.freetheai.xyz addresses and model IDs no longer work. Check the API key separately: current keys start with ftai_, and the error docs define 401 as a missing, incomplete, revoked or expired key.

Current connection values

OpenAI SDK base URL

https://api.freetheai.org/v1

Full Chat Completions URL

https://api.freetheai.org/v1/chat/completions for clients that request the complete endpoint.

Anthropic Messages base URL

https://api.freetheai.org

Model ID

Copy a current ID from the live model list. IDs start with fta/, for example fta/zai/glm-5.3.

API key

Use a complete, active key from Dashboard → API keys. Current keys start with ftai_ and are shown once; if the service returns 401, check that the key is complete and active or create a replacement.

Update the existing app

1. Replace the old API host with the base URL expected by your app. Most OpenAI SDKs take https://api.freetheai.org/v1; do not paste the full /chat/completions path into a base-URL field unless that app explicitly asks for a full endpoint.
2. Check the key in Dashboard → API keys. Current keys start with ftai_; if the old value is missing, incomplete, revoked or expired, create a replacement. Keep it in the app's credential field or a local secret store.
3. Replace the model name with a complete current fta/... ID. Do not remove the prefix or reuse an old .xyz ID.
4. For a Chat Completions client, check that it calls /v1/chat/completions exactly once. A duplicated /v1/v1 or a stale full endpoint can still produce a path error after the host is changed.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.freetheai.org/v1",
    api_key="YOUR_FREETHEAI_KEY",
)

reply = client.chat.completions.create(
    model="fta/zai/glm-5.3",
    messages=[{"role": "user", "content": "Hello"}],
)

This is a configuration example, not a successful live request. Replace the placeholder locally; do not publish a real key.

Check the free-tier gates

FreeTheAI's documentation lists 50 free requests per day after email confirmation, resetting at 00:00 UTC. It also says a daily check-in unlocks free models, free accounts run one request at a time, and linking Discord adds 50 daily requests. These are provider-published limits, not account-tested guarantees.

• 403 with a daily check-in message: finish the human check-in in the FreeTheAI website, then retry.
• 404 model not found: copy a current ID, including its fta/ prefix.
• 429: check daily use and whether another free request is still running; a second concurrent request can be refused.
• 503: the selected model may be busy or paused; the docs say a refused busy request does not count against the daily allowance.

FreeTheAI is a gateway route

FreeTheAI describes the model routes as served by independent sellers, not the original model companies. Its privacy policy says prompts and replies are not stored by FreeTheAI, but it records request metadata and network/device data; the policy says IP is kept as a keyed hash and an encrypted copy. The selected provider receives the message and handles it under its own policy. Check the current route and provider terms before sending sensitive data. A FreeTheAI key is not a direct GLM, Kimi or MiniMax provider key.

Current endpoints and account setup · API key format and errors · 401, 403, 404, 429 and 503 errors · Model IDs and legacy-ID errors · Daily limits and concurrency · Daily check-in · Privacy policy · Current model list

For a side-by-side view of other free routes, see FreeTheAI and other Chinese-model API options or compare access conditions.

Official public documentation checked 2 November 2026. No FreeTheAI account, API key or model request was used; model availability and account eligibility can change.

Free API access and limits

Check the site free API access and limits before trying to obtain a site key. Read the current status and request limits before using a site key. It is not a source of official provider keys.

ChinaGenAI is not affiliated with the providers named here. This release contains setup guidance and explicitly labeled fictional examples, not claimed live model benchmarks.