\r \r

LightningProx — Lightning-Native AI Gateway\r

\r Pay-per-use access to AI models across 5 providers via Bitcoin Lightning micropayments. No API keys. No subscriptions. No accounts. Pay sats, get inference. Implements the L402 payment protocol.\r \r

When to Use\r

\r

• Accessing AI models without provider API keys\r
• Autonomous agent inference with Lightning payments\r
• Comparing responses across multiple providers\r
• Low-cost inference via open models (Llama 4, Mistral, DeepSeek)\r
• Vision tasks (multimodal via image_url)\r
• Code generation (Codestral, Devstral)\r
• Reasoning tasks (Magistral)\r \r

Supported Models\r

\r | Provider | Models (use exact IDs in API calls) |\r |----------|--------------------------------------|\r | Anthropic | claude-opus-4-5-20251101, claude-sonnet-4-20250514, claude-haiku-4-5-20251001 |\r | OpenAI | gpt-4o, gpt-4-turbo |\r | Together.ai | meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8, meta-llama/Llama-3.3-70B-Instruct-Turbo, deepseek-ai/DeepSeek-V3, mistralai/Mixtral-8x7B-Instruct-v0.1 |\r | Mistral | mistral-large-latest, mistral-small-latest, open-mistral-nemo, codestral-latest, devstral-latest, pixtral-large-latest, magistral-medium-latest |\r | Google | gemini-2.5-flash, gemini-2.5-pro |\r \r

Payment Flow\r

\r

Option A — Spend Token (recommended for repeat use)\r

# 1. Top up at lightningprox.com/topup — pay Lightning invoice, get token\r
# 2. Use token directly\r
curl -X POST https://lightningprox.com/v1/messages \\r
  -H "Content-Type: application/json" \\r
  -H "X-Spend-Token: $LIGHTNINGPROX_SPEND_TOKEN" \\r
  -d '{\r
    "model": "claude-opus-4-5-20251101",\r
    "messages": [{"role": "user", "content": "Hello"}],\r
    "max_tokens": 1000\r
  }'\r
```\r
\r
### Option B — L402 Pay-per-request (standard protocol)\r
```bash\r
# 1. Send request without credentials → receive HTTP 402\r
curl -si -X POST https://lightningprox.com/v1/messages \\r
  -H "Content-Type: application/json" \\r
  -d '{"model": "gemini-2.5-flash", "messages": [{"role": "user", "content": "Hello"}], "max_tokens": 100}'\r
# Response header: WWW-Authenticate: L402 macaroon="eyJ...", invoice="lnbc..."\r
\r
# 2. Pay the bolt11 invoice from your Lightning wallet\r
\r
# 3. Retry with L402 credential (macaroon from header + preimage from wallet)\r
curl -X POST https://lightningprox.com/v1/messages \\r
  -H "Content-Type: application/json" \\r
  -H "Authorization: L402 \x3Cmacaroon>:\x3Cpreimage>" \\r
  -d '{"model": "gemini-2.5-flash", "messages": [{"role": "user", "content": "Hello"}], "max_tokens": 100}'\r
```\r
\r
## Drop-in OpenAI SDK Replacement\r
\r
```bash\r
npm install lightningprox-openai\r
```\r
\r
```javascript\r
// Before: import OpenAI from 'openai'\r
import OpenAI from 'lightningprox-openai'\r
const client = new OpenAI({ apiKey: process.env.LIGHTNINGPROX_SPEND_TOKEN })\r
\r
// Everything else stays identical:\r
const response = await client.chat.completions.create({\r
  model: 'claude-opus-4-5-20251101',\r
  messages: [{ role: 'user', content: 'Hello' }]\r
})\r
```\r
\r
Two lines change. Nothing else does.\r
\r
## Check Available Models\r
```bash\r
curl https://lightningprox.com/v1/models\r
# or\r
curl https://lightningprox.com/api/capabilities\r
```\r
\r
## Security Manifest\r
\r
| Permission | Scope | Reason |\r
|------------|-------|--------|\r
| Network | lightningprox.com | API calls for AI inference |\r
| Env Read | LIGHTNINGPROX_SPEND_TOKEN | Authentication for prepaid requests |\r
\r
## Trust Statement\r
\r
LightningProx is operated by LPX Digital Group LLC. Payment = authentication. No data stored beyond request logs. No accounts, no KYC. Operated at lightningprox.com.\r