Chat & Deep Reasoning
Stream responses from Groq, Mistral, NVIDIA, Gemini, Cohere, or Edge GPUs.
Hello! I am your **Universal AI Assistant** connected to 46+ models across NVIDIA, Groq, Mistral, Cohere, Gemini, and Cloudflare Edge GPUs.
Select any model from the dropdown above and start typing!
Text-to-Image Studio
Generate photorealistic 8k artwork in ~2 seconds using Cloudflare Edge GPUs.
No Image Generated Yet
Type a prompt on the left and click Generate to see the magic happen!
Synthesizing pixels on Cloudflare Edge GPUs...
OpenAI Whisper Large v3
Upload any audio file or record your voice to transcribe speech into text instantly.
Drag & Drop Audio File Here
Supports MP3, WAV, M4A, OGG, WEBM
Transcribed Output
Transcription results will appear here in real-time...
BGE Large Vector Embeddings
Generate 1024-dimensional semantic vectors for vector databases, RAG, and cosine similarity search.
Cosine Similarity Score
Raw Vector Preview (First 20 Dimensions)
Click compute to view vector dimensions...
Live Web Intelligence
Perform real-time live internet searches and document reranking through your gateway.
Content Safety & Sentiment Meter
Evaluate prompts against Llama-Guard 3 safety filters and DistilBERT emotion scoring.
Universal AI Gateway — Swagger API & Multi-Platform Hub
Explore OpenAPI-standard interactive documentation for all 59 models across 20 providers. Test endpoints live with auto-failover, or grab copy-paste production SDK code for Android APKs, Web Apps, Desktop Applications, and Chrome Extensions (Manifest V3).
https://tiny-shape-e3d1.mukulgupta0014.workers.dev
System Discovery & Health
2 EndpointsReturns health metadata, currently active provider networks, total models loaded, and edge GPU availability.
Returns an OpenAI-compliant {"object":"list","data":[...]} collection containing all available models across NVIDIA, Groq, Mistral, Gemini, DeepSeek, Cerebras, SambaNova, and Cloudflare Edge GPUs.
Chat Completions & Frontier Reasoning
1 Unified Endpoint (59 Models)OpenAI-compatible chat completion endpoint. Supports all 59 models. Automatically triggers instant zero-downtime failover across Groq, Cerebras, Mistral, GitHub Models, Gemini, NVIDIA NIM Pool, and Cloudflare Edge GPUs if any upstream provider encounters rate-limits.
Multimodal Image & Speech Inference
2 EndpointsGenerates high-definition images using Cloudflare Edge GPUs (FLUX.1 Schnell & SDXL Lightning) with zero-cost Pollinations fallback. Returns raw binary image/jpeg.
Transcribes voice or audio files using Whisper Large v3 Turbo on Cloudflare Edge GPUs. Send raw audio bytes (WAV, MP3, WebM, OGG) directly in the request body with Content-Type: application/octet-stream.
Vector Embeddings, Rerank & Live Search
3 EndpointsComputes 1024-dimensional semantic embeddings for strings or arrays of strings using BAAI BGE-Large on Edge GPUs.
Reranks a list of candidate documents against a query string using Cohere Rerank v3.5, scoring relevance for RAG pipelines.
Performs live internet search and extracts grounded web content snippets with automatic DuckDuckGo fallback.
Translation, Summarization & Moderation
4 EndpointsAndroid APK Integration (Kotlin & OkHttp)
Native Android / APKConnect any Android application (APK) directly to the OmniAI Gateway. Supports streaming Server-Sent Events (SSE) directly into Jetpack Compose or XML Views with automatic token reassembly and background Coroutine execution.
dependencies {
// OkHttp 4 for high-performance HTTP & SSE streaming
implementation("com.squareup.okhttp3:okhttp:4.12.0")
implementation("com.squareup.okhttp3:okhttp-sse:4.12.0")
// Kotlin Coroutines for async execution
implementation("org.jetbrains.kotlinx:kotlinx-coroutines-android:1.8.0")
// JSON Serialization
implementation("com.google.code.gson:gson:2.10.1")
}
<manifest xmlns:android="http://schemas.android.com/apk/res/android">
<!-- Required for OmniAI Gateway Network Access -->
<uses-permission android:name="android.permission.INTERNET" />
<uses-permission android:name="android.permission.ACCESS_NETWORK_STATE" />
</manifest>
package com.example.omniai
import kotlinx.coroutines.Dispatchers
import kotlinx.coroutines.withContext
import okhttp3.*
import okhttp3.MediaType.Companion.toMediaType
import okhttp3.RequestBody.Companion.toRequestBody
import org.json.JSONArray
import org.json.JSONObject
import java.io.BufferedReader
import java.io.InputStreamReader
class OmniAIGatewayClient(
private val baseUrl: String = "https://tiny-shape-e3d1.mukulgupta0014.workers.dev",
private val masterKey: String = "sk_live_master_gateway_key_2026"
) {
private val client = OkHttpClient.Builder().build()
private val jsonMediaType = "application/json; charset=utf-8".toMediaType()
/**
* Streams chat responses token by token into a Kotlin lambda callback
*/
suspend fun streamChat(
model: String = "meta/llama-3.2-11b-vision-instruct",
userPrompt: String,
onTokenReceived: (String) -> Unit,
onComplete: () -> Unit,
onError: (Throwable) -> Unit
) = withContext(Dispatchers.IO) {
val payload = JSONObject().apply {
put("model", model)
put("stream", true)
put("messages", JSONArray().apply {
put(JSONObject().put("role", "system").put("content", "You are an intelligent Android assistant."))
put(JSONObject().put("role", "user").put("content", userPrompt))
})
}
val request = Request.Builder()
.url("$baseUrl/v1/chat/completions")
.addHeader("Authorization", "Bearer $masterKey")
.post(payload.toString().toRequestBody(jsonMediaType))
.build()
try {
val response = client.newCall(request).execute()
if (!response.isSuccessful) throw Exception("HTTP Error: ${response.code}")
val reader = BufferedReader(InputStreamReader(response.body?.byteStream() ?: return@withContext))
var line: String?
while (reader.readLine().also { line = it } != null) {
val currentLine = line?.trim() ?: continue
if (currentLine.startsWith("data:")) {
val data = currentLine.removePrefix("data:").trim()
if (data == "[DONE]") break
try {
val json = JSONObject(data)
val content = json.getJSONArray("choices")
.getJSONObject(0)
.getJSONObject("delta")
.optString("content", "")
if (content.isNotEmpty()) {
withContext(Dispatchers.Main) { onTokenReceived(content) }
}
} catch (e: Exception) { /* skip partial malformed packets */ }
}
}
withContext(Dispatchers.Main) { onComplete() }
} catch (t: Throwable) {
withContext(Dispatchers.Main) { onError(t) }
}
}
}
Web Integration (React, Next.js & Vanilla JS)
Web & Mobile Web
Easily integrate the OmniAI Gateway into React, Next.js, Vue, or Vanilla HTML/JavaScript using standard standard browser Fetch and the ReadableStream API.
import { useState, useCallback } from 'react';
const GATEWAY_URL = "https://tiny-shape-e3d1.mukulgupta0014.workers.dev";
const MASTER_KEY = "sk_live_master_gateway_key_2026";
export function useOmniAI() {
const [response, setResponse] = useState("");
const [isGenerating, setIsGenerating] = useState(false);
const [error, setError] = useState<string | null>(null);
const askModel = useCallback(async (prompt: string, model = "meta/llama-3.2-11b-vision-instruct") => {
setIsGenerating(true);
setResponse("");
setError(null);
try {
const res = await fetch(`${GATEWAY_URL}/v1/chat/completions`, {
method: "POST",
headers: {
"Content-Type": "application/json",
"Authorization": `Bearer ${MASTER_KEY}`
},
body: JSON.stringify({
model,
messages: [{ role: "user", content: prompt }],
stream: true
})
});
if (!res.ok) throw new Error(`Gateway returned HTTP ${res.status}`);
const reader = res.body?.getReader();
if (!reader) throw new Error("No readable stream");
const decoder = new TextDecoder();
let accumulated = "";
while (true) {
const { done, value } = await reader.read();
if (done) break;
const chunk = decoder.decode(value, { stream: true });
const lines = chunk.split("\n");
for (const line of lines) {
const trimmed = line.trim();
if (trimmed.startsWith("data:") && !trimmed.includes("[DONE]")) {
try {
const data = JSON.parse(trimmed.replace(/^data:\s*/, ''));
const delta = data.choices?.[0]?.delta?.content || "";
accumulated += delta;
setResponse(accumulated);
} catch (e) {}
}
}
}
} catch (err: any) {
setError(err.message);
} finally {
setIsGenerating(false);
}
}, []);
return { askModel, response, isGenerating, error };
}
async function generateAIResponse(prompt, onToken) {
const res = await fetch("https://tiny-shape-e3d1.mukulgupta0014.workers.dev/v1/chat/completions", {
method: "POST",
headers: {
"Content-Type": "application/json",
"Authorization": "Bearer sk_live_master_gateway_key_2026"
},
body: JSON.stringify({
model: "meta/llama-3.2-11b-vision-instruct",
messages: [{ role: "user", content: prompt }],
stream: true
})
});
const reader = res.body.getReader();
const decoder = new TextDecoder();
while (true) {
const { done, value } = await reader.read();
if (done) break;
const chunk = decoder.decode(value);
for (const line of chunk.split("\n")) {
if (line.startsWith("data:") && !line.includes("[DONE]")) {
try {
const content = JSON.parse(line.slice(5)).choices[0].delta.content;
if (content) onToken(content);
} catch (e) {}
}
}
}
}
Desktop App Integration (Python, C# .NET, Electron)
Windows • macOS • Linux
Because the OmniAI Gateway implements standard OpenAI REST & SSE interfaces, you can drop it directly into official OpenAI Python SDKs, C# HttpClient, or Electron Desktop apps by merely pointing base_url to the gateway!
import os
from openai import OpenAI
# 1. Initialize client pointing directly to your OmniAI Gateway
client = OpenAI(
base_url="https://tiny-shape-e3d1.mukulgupta0014.workers.dev/v1",
api_key="sk_live_master_gateway_key_2026"
)
# 2. Call any of the 59 models (e.g. DeepSeek R1, Groq, Codestral, Gemini 2.0 Flash)
response = client.chat.completions.create(
model="deepseek-r1", # or "cerebras/llama-3.3-70b", "codestral", "gemini-2.0-flash"
messages=[
{"role": "system", "content": "You are a master desktop automation copilot."},
{"role": "user", "content": "Generate a Python script to monitor clipboard history with timestamp."}
],
stream=True
)
# 3. Stream output tokens to terminal or desktop UI
for chunk in response:
content = chunk.choices[0].delta.content or ""
print(content, end="", flush=True)
using System;
using System.Net.Http;
using System.Net.Http.Headers;
using System.Text;
using System.Text.Json;
using System.Threading.Tasks;
public class OmniAIDesktopClient
{
private static readonly HttpClient client = new HttpClient();
private const string GatewayUrl = "https://tiny-shape-e3d1.mukulgupta0014.workers.dev/v1/chat/completions";
private const string MasterKey = "sk_live_master_gateway_key_2026";
public static async Task<string> AskAIAsync(string prompt, string model = "meta/llama-3.2-11b-vision-instruct")
{
var requestBody = new
{
model = model,
messages = new[] { new { role = "user", content = prompt } },
stream = false
};
var request = new HttpRequestMessage(HttpMethod.Post, GatewayUrl);
request.Headers.Authorization = new AuthenticationHeaderValue("Bearer", MasterKey);
request.Content = new StringContent(JsonSerializer.Serialize(requestBody), Encoding.UTF8, "application/json");
var response = await client.SendAsync(request);
response.EnsureSuccessStatusCode();
var json = await response.Content.ReadAsStringAsync();
using var doc = JsonDocument.Parse(json);
return doc.RootElement.GetProperty("choices")[0].GetProperty("message").GetProperty("content").GetString();
}
}
Chrome Extension Manifest V3 Integration
Manifest V3 Extension
In Manifest V3, background service workers are ephemeral and cannot use simple WebSockets or standard long-lived intervals without proper handling. This architecture uses chrome.runtime.connect long-lived ports to deliver low-latency token streaming to popup and sidepanel windows.
{
"manifest_version": 3,
"name": "ClipIQ AI Assistant",
"version": "1.0.0",
"permissions": [
"storage",
"clipboardRead"
],
"host_permissions": [
"https://*.workers.dev/*",
"https://tiny-shape-e3d1.mukulgupta0014.workers.dev/*"
],
"background": {
"service_worker": "background.js",
"type": "module"
},
"action": {
"default_popup": "popup.html"
}
}
const GATEWAY_URL = "https://tiny-shape-e3d1.mukulgupta0014.workers.dev";
const MASTER_KEY = "sk_live_master_gateway_key_2026";
// Listen for connection from popup or sidebar
chrome.runtime.onConnect.addListener((port) => {
if (port.name !== "omniai-stream") return;
port.onMessage.addListener(async (msg) => {
if (msg.type === "GENERATE_AI") {
try {
const res = await fetch(`${GATEWAY_URL}/v1/chat/completions`, {
method: "POST",
headers: {
"Content-Type": "application/json",
"Authorization": `Bearer ${MASTER_KEY}`
},
body: JSON.stringify({
model: msg.model || "meta/llama-3.2-11b-vision-instruct",
messages: msg.messages,
stream: true
})
});
if (!res.ok) {
port.postMessage({ type: "ERROR", error: `HTTP ${res.status}` });
return;
}
const reader = res.body.getReader();
const decoder = new TextDecoder();
while (true) {
const { done, value } = await reader.read();
if (done) break;
const chunk = decoder.decode(value);
const lines = chunk.split("\n");
for (const line of lines) {
const trimmed = line.trim();
if (trimmed.startsWith("data:") && !trimmed.includes("[DONE]")) {
try {
const data = JSON.parse(trimmed.replace("data:", "").trim());
const token = data.choices?.[0]?.delta?.content || "";
if (token) port.postMessage({ type: "TOKEN", token });
} catch (e) {}
}
}
}
port.postMessage({ type: "DONE" });
} catch (err) {
port.postMessage({ type: "ERROR", error: err.message });
}
}
});
});
// Connect long-lived port to background service worker
const port = chrome.runtime.connect({ name: "omniai-stream" });
port.onMessage.addListener((msg) => {
if (msg.type === "TOKEN") {
document.getElementById("output").textContent += msg.token;
} else if (msg.type === "DONE") {
console.log("Streaming finished");
} else if (msg.type === "ERROR") {
alert("Error: " + msg.error);
}
});
// Trigger generation
function sendPrompt(userText) {
document.getElementById("output").textContent = "";
port.postMessage({
type: "GENERATE_AI",
model: "meta/llama-3.2-11b-vision-instruct",
messages: [{ role: "user", content: userText }]
});
}
Gateway Authentication Architecture
All endpoints (except /health and /v1/models) are authenticated using Master Gateway Tokens. The gateway supports two interchangeable authorization schemes:
Authorization: Bearer <MASTER_TOKEN>
Standard RFC 6750 bearer header. Compatible with official OpenAI, Anthropic, and LangChain client libraries.
X-API-Key: <MASTER_TOKEN>
Custom API key header for environments where overriding Authorization headers is restricted.
Pre-configured Master Tokens in Active Gateway
| Token | Platform Audience | Privilege | Rate Limits |
|---|---|---|---|
sk_live_master_gateway_key_2026 |
Studio Web Playground & Admin | Master Admin | Unlimited (Self-Failover) |
sk_live_web_client_key_9a8b7c |
Production Web Integrations | Web Client | Unlimited (Auto-Tier) |
sk_live_mobile_app_key_4d5e6f |
Android APK & Mobile Apps | Mobile SDK | Unlimited (Auto-Tier) |
4-Tier Zero-Downtime Auto-Failover Flow
The gateway guarantees near 100% uptime through instant cascading failover across 20 global provider clusters:
Primary Requested Model
Direct low-latency route to Groq, Cerebras, Mistral, Cohere, Gemini, SambaNova, or Perplexity.
If HTTP 429 / 5xx / Rate Limit ↓Cloudflare Native Edge GPUs
Immediate failover to Cloudflare Workers AI edge GPUs (Llama 3.1 8B, Qwen 2.5 7B, Mistral 7B) with zero rate limits.
If Edge Queue Saturated ↓NVIDIA NIM 14-Key Pool
Rotates round-robin across 14 independent high-throughput NVIDIA NIM API keys (Llama 3.2 Vision, Gemma 4, Minimax).
If All Enterprise Pools Busy ↓Pollinations Decentralized Cluster
Final zero-key decentralized AI cluster (OpenAI & Mistral instances) ensuring continuous unbroken output.
HTTP Status Code Matrix
| Status | Meaning | Gateway Behavior |
|---|---|---|
| 200 OK | Success | Inference returned or SSE stream opened successfully. |
| 400 Bad Request | Malformed JSON | Invalid parameters or unsupported model specified. |
| 401 Unauthorized | Invalid Master Key | Authorization header is missing or does not match master tokens. |
| 503 Networks Busy | All Tiers Exhausted | Triggered only if all 4 cascading fallback tiers fail simultaneously. |