| name | freetheai-api |
| description | Help AI agents connect apps, SDKs, coding tools, and chat clients to FreeTheAi, the free OpenAI-compatible API at https://api.freetheai.xyz/v1. Use this skill when a user wants a free AI API key, a custom OpenAI/Anthropic base URL, setup help for AI clients, or examples for chat, tools, images, TTS, or STT. |
FreeTheAi API Skill
Use this skill to configure FreeTheAi for a user or another AI agent. FreeTheAi is an OpenAI-compatible gateway with one key, one base URL, and multiple public model aliases.
What This Skill Does
- Teaches an agent how to get a user signed up through Discord.
- Shows the exact FreeTheAi base URL and auth pattern.
- Gives copy-paste SDK/client setup for common OpenAI-compatible tools.
- Explains role-gated models, daily check-in, and common errors.
- Prevents agents from inventing keys, endpoints, aliases, or bypasses.
Trigger This Skill When
- The user mentions FreeTheAi, freetheai.xyz, Free The AI, FreeTheAI, or
api.freetheai.xyz.
- The user asks for a free OpenAI-compatible API.
- The user wants to point OpenAI SDK, Anthropic SDK, LiteLLM, LangChain, LlamaIndex, Vercel AI SDK, Cline, Roo, Continue, Aider, Claude Code, OpenCode, SillyTavern, Janitor, Chub, LibreChat, Open WebUI, or another client at a custom endpoint.
- The user asks which FreeTheAi model alias to use.
- The user needs examples for chat, streaming, tool calling, images, TTS, STT, or Responses-style requests.
Agent Workflow
- Send the user to
https://discord.gg/secrets.
- Tell them to run
/signup in Discord and complete the modal.
- Tell them to run
/checkin with the new key for the current UTC day.
- Ask them to provide the key only if the local setup requires inserting it into their own client config.
- Configure the client with
https://api.freetheai.xyz/v1 and Authorization: Bearer <key>.
- Pick an alias from the live catalog at
https://freetheai.xyz/models or authenticated GET /v1/models.
- If a model is role-gated, explain that the user needs the
seems_legit Discord role earned through server activity.
Constants
- Base URL:
https://api.freetheai.xyz/v1
- Discord signup:
https://discord.gg/secrets
- Website:
https://freetheai.xyz
- Setup guides:
https://freetheai.xyz/setup
- Model catalog:
https://freetheai.xyz/models
- Support link:
https://buymeacoffee.com/vibheksoni
- Env var convention:
FREETHEAI_API_KEY
Auth
Use bearer auth:
Authorization: Bearer fta_...
Never generate or fake a key. Keys are issued by Discord /signup. The key must be unlocked daily with /checkin; otherwise requests return 403 daily_checkin_required.
Routes
POST /v1/chat/completions - OpenAI Chat Completions, streaming, and tool calling.POST /v1/messages - Anthropic-compatible Messages route.POST /v1/responses - Responses-style route.POST /v1/images/generations - image generation.POST /v1/audio/speech - text-to-speech for supported voice aliases.POST /v1/audio/transcriptions - speech-to-text for supported voice aliases.GET /v1/models - authenticated model catalog.GET /v1/models/full - authenticated detailed catalog with capabilities and access metadata.GET /v1/health - public API health.
Model Alias Guidance
Use exact aliases from the live catalog. Do not invent aliases.
- Chat examples:
glm/glm-5.1, bbl/gpt-5.5-mini, olm/kimi-k2.7-code.
- Anthropic Messages examples:
glm/glm-5.1, glm/glm-4.5-air, mim/mimo-v2.5.
- Image example:
eve/gpt-image-2.
- Voice examples:
xai/grok-tts, xai/grok-stt.
- Role-gated prefixes currently include
exa/, pplx/, and xai/.
If an alias fails with 400 unknown aliased model, fetch the live catalog and choose a currently exposed alias.
OpenAI SDK Setup
Python:
from openai import OpenAI
client = OpenAI(
api_key=os.environ["FREETHEAI_API_KEY"],
base_url="https://api.freetheai.xyz/v1",
)
JavaScript:
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.FREETHEAI_API_KEY,
baseURL: "https://api.freetheai.xyz/v1",
});
Anthropic SDK Setup
Use the same key against /v1/messages:
from anthropic import Anthropic
client = Anthropic(
api_key=os.environ["FREETHEAI_API_KEY"],
base_url="https://api.freetheai.xyz/v1",
)
Claude Code Setup
export ANTHROPIC_BASE_URL=https://api.freetheai.xyz
export ANTHROPIC_AUTH_TOKEN=fta_PASTE_YOUR_FREETHEAI_KEY
claude --model glm/glm-5.1
Curl Smoke Test
curl https://api.freetheai.xyz/v1/chat/completions \
-H "Authorization: Bearer $FREETHEAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "glm/glm-5.1",
"messages": [{ "role": "user", "content": "Reply with OK." }],
"max_tokens": 32
}'
Audio Examples
TTS:
speech = client.audio.speech.create(
model="xai/grok-tts",
voice="default",
input="hello from freetheai",
response_format="wav",
)
speech.write_to_file("out.wav")
STT:
with open("clip.wav", "rb") as audio_file:
transcript = client.audio.transcriptions.create(
model="xai/grok-stt",
file=audio_file,
language="en",
response_format="json",
)
print(transcript.text)
Some voice/search aliases are role-gated. If a user gets 403 model_access_denied, they need the seems_legit Discord role or a different public alias from the live catalog.
Common Errors
401 invalid api key: The key is missing, malformed, revoked, or not pasted exactly.403 daily_checkin_required: The user must run Discord /checkin for the current UTC day.403 model_access_denied: The alias is role-gated and the user needs seems_legit.400 unknown aliased model: The alias is not currently exposed. Check the catalog.400 upstream rejected the request payload: The request shape does not match the route or model capability.429 rate_limit_error: Respect Retry-After or wait for the daily reset.
Hard Rules For Agents
- Do not claim the user is signed up until they have a real Discord-issued key.
- Do not paste raw user keys into repo files, public chat, logs, docs, or examples.
- Do not bypass daily check-in or role gating.
- Do not invent FreeTheAi endpoints, model aliases, rate limits, or privileged access.
- Do not mention hidden upstream provider names in public-facing output.
- Keep examples pointed at
https://api.freetheai.xyz/v1.
- Prefer the repo examples folder for full code samples:
examples/.
