Synaps

---

Why Synaps

Most AI apps ask for a monthly subscription to one company's models. Synaps works the other way round:

• Every major model in one place. Switch between Claude, GPT, Gemini, Grok, DeepSeek, Qwen, Kimi, GLM, Mistral, Llama and image models mid-conversation.
• Pay only for what you use. Top up from $1 and each message is deducted at the provider's price plus a flat, visible fee. Prices are shown before you send.
• Wallet, not account. Sign in by signing a message with your wallet. No email, no password, no card.
• One key for builders. The same balance powers an OpenAI-compatible API, so any OpenAI SDK works by changing the base URL.
• Clear privacy labels. Every model shows who can read your prompts: Free, No training, or Provider can read.

Features

💬 Chat	Streaming replies with Markdown and code blocks, model name and cost on every reply, synced history per wallet
🌐 Web search	One toggle; the model reads live results and cites numbered sources
📎 Attachments	Photos, PDFs and text/code files, with a capability check and a one-tap switch to a model that can handle them
🎨 Image generation	Image mode for models that create pictures, with download
🧭 Custom instructions	Default language, tone and context applied to every reply, synced across devices
🔗 Share links	Read-only public snapshot of a chat, with Stop sharing
🔑 API keys	OpenAI-compatible /api/v1, per-key daily spend limits, streaming, tool calling, JSON mode
💸 Top-ups	ETH on Robinhood Chain or Base, USDC on Base, verified on-chain and credited once per transaction
🎁 Referrals	Invite link on the Wallet page; both sides get bonus credit on the friend's first top-up
🪙 $SYNAPS ready	The token is detected automatically when the treasury launches it on Pons, then becomes a top-up option with bonus credit
📱 Installable	PWA with offline screen and app shortcuts on Android, iOS and desktop
📊 Admin dashboard	OpenRouter credit and runway, treasury balances, revenue and profit, top models, alerts

Use the API

Create a key on the API keys page, then point any OpenAI client at Synaps.

curl https://www.synaps.chat/api/v1/chat/completions \
  -H "Authorization: Bearer $SYNAPS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-sonnet-5.5",
    "messages": [{ "role": "user", "content": "Hello!" }]
  }'

from openai import OpenAI

client = OpenAI(base_url="https://www.synaps.chat/api/v1", api_key="syn_...")

reply = client.chat.completions.create(
    model="anthropic/claude-sonnet-5.5",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(reply.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://www.synaps.chat/api/v1", apiKey: process.env.SYNAPS_API_KEY });

const stream = await client.chat.completions.create({
  model: "google/gemini-3.8-flash",
  messages: [{ role: "user", content: "Write a haiku about block times." }],
  stream: true,
});
for await (const chunk of stream) process.stdout.write(chunk.choices[0]?.delta?.content ?? "");

GET /api/v1/models lists every model with its live price, context length and input/output modalities. Every response includes usage.cost, the amount charged in USD. Full reference: synaps.chat/docs.

How it works

flowchart LR
  W[Wallet] -- sign message --> S[Synaps]
  W -- ETH / USDC / $SYNAPS --> T[(Treasury)]
  S -- verify receipt on-chain --> C[Robinhood Chain / Base]
  S -- credit USD balance --> DB[(Supabase)]
  U[Chat or API key] --> S
  S -- route request --> OR[OpenRouter] --> M[Claude · GPT · Gemini · …]
  M -- reply + usage.cost --> S
  S -- debit cost × (1 + fee) --> DB

1. Sign in. The wallet signs a one-time nonce (no gas). The server verifies it and sets an HTTP-only session cookie.
2. Top up. The wallet sends ETH, USDC or $SYNAPS to the treasury. The server reads the receipt on-chain, checks sender, recipient and amount, and credits the USD value once per transaction hash.
3. Use a model. Requests are routed through OpenRouter. Paid requests to open-weight models are restricted to providers that do not store or train on prompts.
4. Pay for it. After each reply the provider's reported cost × (1 + fee) is debited from the balance, even if the client disconnects mid-stream.

All balances are stored in micro-dollars (USD × 10⁶) and changed only inside database functions, so credits and debits are atomic.

Tech stack

• App: Next.js 15 (App Router), React 19, TypeScript
• Design: custom CSS design system, three.js, GSAP ScrollTrigger and Lenis on the landing page, Geist and Instrument Serif
• Wallets: Privy for connection, viem for signature checks, receipts, ERC-20 transfers and pool prices
• Data: Supabase Postgres, with row-level security on every table and server-only access through a service role
• Models: OpenRouter, with fallbacks for free models, PDF parsing and the web-search plugin
• Hosting: Vercel, with Web Analytics and the GitHub integration

Run it yourself

Requirements: Node.js 20+, a Supabase project, an OpenRouter key and a Privy app.

git clone https://github.com/onsynaps/Synaps-.git synaps
cd synaps
npm install
cp .env.example .env.local   # fill in the required values

1. Open the Supabase SQL editor and run supabase/schema.sql. It creates the tables and functions and locks them to the service role.
2. In the Privy dashboard, add http://localhost:3000 (and your domain) to Allowed domains.
3. Start the dev server:

npm run dev

The landing page is at http://localhost:3000, the app at /app, the docs at /docs and the admin dashboard at /app/admin (for the wallets in ADMIN_ADDRESSES).

Environment variables

Variable	Required	Purpose
OPENROUTER_API_KEY	✓	Routes requests to model providers
SUPABASE_URL / SUPABASE_SERVICE_ROLE_KEY	✓	Database (server only)
SESSION_SECRET	✓	Signs session cookies
NEXT_PUBLIC_TREASURY	✓	Wallet that receives top-ups
NEXT_PUBLIC_PRIVY_APP_ID	✓	Wallet connection
NEXT_PUBLIC_SITE_URL	✓	Canonical URL for links, sitemap and the API base
MARKUP		Fee on top of provider cost (default 0.2)
FREE_DAILY_LIMIT / FREE_GLOBAL_DAILY_LIMIT		Free-model allowance per visitor and in total
WEB_SEARCH_FEE_USD		Flat fee per web-search message
REF_*		Referral percentages, cap and minimum
SYNAPS_*		$SYNAPS bonus and manual token or pool override
ADMIN_ADDRESSES, LOW_CREDIT_USD		Admin access and the credit alert level
RH_RPC_URL, BASE_RPC_URL		RPC endpoints (a private RPC is recommended in production)

See .env.example for every option and its default.

Project structure

app/
  page.tsx               landing page
  app/                   the app: chat, wallet, API keys, models, admin
  api/                   chat, auth, top-ups, keys, chats, share, referral, token, admin
  api/v1/                OpenAI-compatible API (chat/completions, models)
  models/                public, indexable model pages
  docs/                  documentation
  s/[id]/                shared chats
components/              UI: landing, app shell, docs kit, logo, PWA
lib/                     billing relay, models, sessions, chains, pricing, token detection
supabase/schema.sql      database schema and functions
public/                  icons and service worker

Security

• The service-role key and OpenRouter key are used only on the server. The browser never sees them.
• Sessions are HTTP-only cookies created from a verified wallet signature over a single-use nonce.
• API keys are stored as SHA-256 hashes. The full key is shown once, when it is created.
• Top-ups are credited only after the transaction is confirmed on-chain, from the signed-in wallet, to the treasury, and only once per transaction hash.
• $SYNAPS is recognised only if the Pons factory reports it was launched by the treasury wallet, in a transaction the treasury sent itself.
• Prompts sent through the API are never stored. Chat history is stored only for signed-in users with sync on, and never includes file contents or generated images.

Found a vulnerability? Please report it privately by DM to @onsynaps rather than opening a public issue.

Links

• App: synaps.chat/app
• Models & prices: synaps.chat/models
• Docs: synaps.chat/docs
• X: @onsynaps