Chat with an open model, watch an agent call tools, remove backgrounds, generate images, transcribe speech and run OCR, all on one page. Embedded tools load only when you open their tab.
Chat with an open model
Ask a coding or product question. Answers come from an open-weight model on Groq’s free tier through our own PHP proxy; your key is never needed.
Coming soon. The live chat demo switches on once our free model key is connected. Meanwhile, try the tools in the other tabs.
Runs on an open model through our server (Groq free tier). Don’t enter personal data. Fair-use limits per visitor per hour and per day, plus a daily cap for the whole site.
Gradio on CPU (likely calls a hosted Qwen API) · Space licence apache-2.0 · Model Qwen3 0.6B–14B+ (apache-2.0). Official Qwen space. Third-party Space by Qwen, checked running on 17 Oct 2026.
Agent demo: an AI that calls tools
The agent can call three read-only tools on our server: search_aiuseflow_tools (searches this site), get_free_api_limits (reads our checked free-API data) and calculator. Each tool call is shown as a step, so you can see what an agent framework does under the hood.
Coming soon. The live agent demo switches on once our free model key is connected. Meanwhile, try the tools in the other tabs.
Runs on an open model through our server (Groq free tier). Don’t enter personal data. Fair-use limits per visitor per hour and per day, plus a daily cap for the whole site.
Image tools
Background removal, image generation and upscaling. Hosted on Hugging Face Spaces and embedded here.
static (WebGPU, runs on visitor device) · Space licence apache-2.0 · Model Xenova/modnet (apache-2.0). No server quota; needs a WebGPU browser (recent Chrome/Edge); first run downloads the model. Third-party Space by webml-community, checked running on 17 Oct 2026.
Gradio on ZeroGPU · Space licence mit · Model ZhengPeng7/BiRefNet (mit). ZeroGPU quota: anonymous visitors get 2 min GPU/day, free HF accounts 5 min. Runs on Hugging Face ZeroGPU: visitors who are not logged in to Hugging Face get about 2 minutes of GPU time per day, so you may see a quota message. Log in to Hugging Face for more. Third-party Space by not-lain, checked running on 17 Oct 2026.
Gradio on ZeroGPU · Space licence mit (space card) · Model black-forest-labs/FLUX.1-schnell (apache-2.0). ZeroGPU quota as above; official BFL space. Runs on Hugging Face ZeroGPU: visitors who are not logged in to Hugging Face get about 2 minutes of GPU time per day, so you may see a quota message. Log in to Hugging Face for more. Third-party Space by black-forest-labs, checked running on 17 Oct 2026.
Gradio on ZeroGPU · Space licence apache-2.0 · Model Real-ESRGAN + SD tile (juggernaut_reborn: mit). ZeroGPU quota; heavy. Runs on Hugging Face ZeroGPU: visitors who are not logged in to Hugging Face get about 2 minutes of GPU time per day, so you may see a quota message. Log in to Hugging Face for more. Third-party Space by gokaygokay, checked running on 17 Oct 2026.
Voice: text-to-speech and speech-to-text
Some of these run fully in your browser (WebGPU), so nothing leaves your device. The first run downloads the model.
static (runs on visitor device) · Space licence apache-2.0 · Model onnx-community/Kokoro-82M-ONNX (apache-2.0). No server quota. Third-party Space by webml-community, checked running on 17 Oct 2026.
Gradio on CPU · Space licence apache-2.0 · Model Kokoro/VITS models via sherpa-onnx (some model cards list no license). CPU space: slower under load but no GPU quota. Third-party Space by k2-fsa, checked running on 17 Oct 2026.
Gradio on CPU (no models listed; likely calls a hosted Qwen API backend) · Space licence apache-2.0 · Model unknown (backend not listed). Depends on Qwen-hosted backend availability. Third-party Space by Qwen, checked running on 17 Oct 2026.
static (runs on visitor device) · Space licence none on space card · Model Xenova/whisper-* (apache-2.0). No server quota; space card has no license field (code is transformers.js example). Third-party Space by Xenova, checked running on 17 Oct 2026.
static (WebGPU) · Space licence mit · Model CohereLabs/cohere-transcribe-03-2026 (apache-2.0, gated=auto). Needs WebGPU. Third-party Space by CohereLabs, checked running on 17 Oct 2026.
OCR: image or PDF to text
Extract text or markdown from images and scanned documents.
Gradio on CPU · Space licence apache-2.0 · Model PP-OCRv5 (apache-2.0, per space). Official PaddlePaddle space. Third-party Space by PaddlePaddle, checked running on 17 Oct 2026.
Gradio on ZeroGPU · Space licence mit · Model deepseek-ai/DeepSeek-OCR-2 (apache-2.0). ZeroGPU quota. Runs on Hugging Face ZeroGPU: visitors who are not logged in to Hugging Face get about 2 minutes of GPU time per day, so you may see a quota message. Log in to Hugging Face for more. Third-party Space by merterbak, checked running on 17 Oct 2026.
Prompt-to-web-page builder
Describe a page and get working HTML. A quick way to feel what vibe coding is like.
Gradio on CPU (likely calls a hosted Qwen API) · Space licence apache-2.0 · Model unknown (backend not listed). Strong fit for vibe coders ("build a landing page" demo). Third-party Space by Qwen, checked running on 17 Oct 2026.
Code snippets: agent SDK quickstarts
These SDKs need Python or Node, so they cannot run on this page. Copy the official quickstart and run it locally.
OpenAI Agents SDK (Python)
Install
pip install openai-agents
Quickstart (python)
from agents import Agent, Runner
agent = Agent(name="Assistant", instructions="You are a helpful assistant")
result = Runner.run_sync(agent, "Write a haiku about recursion in programming.")
print(result.final_output)
import { generateText } from 'ai';
const { text } = await generateText({
model: 'openai/gpt-6-astra', // use Vercel AI Gateway
prompt: 'What is an agent?',
});
Needs: AI_GATEWAY_API_KEY (Vercel AI Gateway; $5/month free credits) · Source
smolagents
Install
pip install "smolagents[toolkit]"
Quickstart (python)
from smolagents import CodeAgent, WebSearchTool, InferenceClientModel
model = InferenceClientModel()
agent = CodeAgent(tools=[WebSearchTool()], model=model, stream_outputs=True)
agent.run("How many seconds would it take for a leopard at full speed to run through Pont des Arts?")
Needs: HF token (Inference Providers are paid for free accounts) · Source