• POST /v1/chat/completions - chat with streaming and tool calling
• POST /v1/messages - Anthropic-style messages route
• POST /v1/responses - OpenAI Responses API
• POST /v1/images/generations - image generation
• POST /v1/images/edits - image edits for supported image aliases
• POST /v1/audio/speech and /v1/audio/transcriptions - supported voice aliases
• Tool calling, structured outputs, multi-turn conversations
• No billing, no credit card, no prompt storage
• Optional paid slots at freetheai.xyz/pricing for separate higher-power models

Quick Start

Base URL    https://api.freetheai.xyz/v1
Auth        Bearer YOUR_API_KEY

1 - Join Discord

Open discord.gg/secrets

2 - Get a key

Run /signup and complete the Discord modal

3 - Check in

Run /checkin with your API key once per UTC day

4 - Build

Point any OpenAI SDK at the base URL

[!IMPORTANT] New keys must complete /checkin once per UTC day before API use. /checkin asks for your existing key and a randomized human challenge.

[!TIP] Lost your key? Run /resetkey and provide a real reset reason - same account, same stats, fresh key.

Setup Guides

Step-by-step setup for the most common AI clients lives at freetheai.xyz/setup.

Each guide ships exact field labels, a copy-paste config, and the gotchas users actually hit.

SKILL.md

This repo includes a SKILL.md describing how AI agents should connect to FreeTheAi (base URL, routes, model aliases, capability metadata, and rate limits). The standalone skeuomorphic UI skill that previously lived here is now at docs/SKILL_SKEUOMORPHIC.md.

Routes

Code Examples

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": "Write a Python hello world." }
    ],
    "stream": true
  }'

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.FREETHEAI_API_KEY,
  baseURL: "https://api.freetheai.xyz/v1",
});

const res = await client.chat.completions.create({
  model: "glm/glm-5.1",
  messages: [{ role: "user", content: "Say hello." }],
});

console.log(res.choices[0].message.content);

from openai import OpenAI

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

res = client.chat.completions.create(
    model="glm/glm-5.1",
    messages=[{"role": "user", "content": "Say hello."}],
)

print(res.choices[0].message.content)

curl https://api.freetheai.xyz/v1/messages \
  -H "Authorization: Bearer $FREETHEAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm/glm-5.1",
    "max_tokens": 256,
    "messages": [
      { "role": "user", "content": "Write a migration plan." }
    ]
  }'

from openai import OpenAI

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

res = client.chat.completions.create(
    model="glm/glm-5.1",
    tool_choice="required",
    messages=[{"role": "user", "content": "Get weather for Boston."}],
    tools=[{
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get the weather for a city.",
            "parameters": {
                "type": "object",
                "properties": {"city": {"type": "string"}},
                "required": ["city"],
                "additionalProperties": False,
            },
        },
    }],
)

print(res.choices[0].message.tool_calls)

curl https://api.freetheai.xyz/v1/images/generations \
  -H "Authorization: Bearer $FREETHEAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "eve/gpt-image-2",
    "prompt": "A neon sports car under rainy city lights"
  }'

The free API exposes eve/gpt-image-2, eve/gpt-image-2-low, and eve/gpt-image-2-medium for image generation. It also exposes ever/*image* aliases for image generation and supported image edits. Live availability and any future image models are in GET /v1/models.

Image responses are OpenAI-compatible and return:

• data[0].b64_json for base64 image data
• For long EVE jobs, send background: true or async: true, then poll GET /v1/images/generations/{request_id} with the same API key.

See examples/image_client.py for a beginner-friendly CLI tool that saves the key locally and writes the returned b64_json image to disk.

Release And Growth

Public releases make updates easier to follow and help the repo look active to GitHub users.

• Latest release draft notes: RELEASE_NOTES.md
• Public changelog: CHANGELOG.md
• SEO and GitHub growth checklist: docs/GROWTH.md

If this project helps you, star the repo and share the docs page. It helps more developers find a free OpenAI-compatible API without a billing setup.

Support The Project

FreeTheAi is built and maintained by Vibhek Soni. The free tier always stays free. If this project saves you a subscription, replaces a paid plan, or unblocks something you were stuck on, a small tip helps cover servers, proxies, and the long nights spent keeping models alive.

• One-time tip on Buy Me a Coffee: buymeacoffee.com/vibheksoni
• Star the repo: github.com/Free-The-Ai/free-ai
• Share the docs and the setup guides with anyone you know who needs a free OpenAI-compatible key

Thanks for keeping the free tier alive.

Models

Browse the full searchable catalog at freetheai.xyz/models. Request examples and endpoint details are on freetheai.xyz/docs.

Live health currently reports 64 public catalog models across 12 provider prefixes. The bundled site fallback snapshot at site/public/models.json is refreshed from the same live /v1/models/full catalog used by the models page.

[!NOTE] Use exact alias IDs from GET /v1/models. Model availability updates automatically as the service catalog changes.

Rate Limits

Tiers unlock through Discord invite progress. Per-user daily success cap defaults to 250/day (resets at UTC midnight) with role-based bumps for trusted members.

Privacy

FreeTheAi does not store prompt text, completion text, uploaded media, or conversation history as operational logs.

Tracked

• IP address and network/security signals
• API key prefix and account/user identifiers
• Discord eligibility, role, and check-in state
• Model alias used
• Route, timestamp, latency, and token/request-unit counts
• Request timestamp
• Status code
• Error IDs and rate-limit/concurrency signals

Not tracked

• Prompt content
• Completion content
• Uploaded media as message logs
• Conversation history

See the