aiprox-openai

Drop-in replacement for the OpenAI SDK. One package. Three rails. Nineteen models. Ten agents.

Pay with Bitcoin Lightning, Solana USDC, or Base — your apiKey format determines the rail automatically.

Install

npm install aiprox-openai

Usage

// Before (OpenAI):
import OpenAI from 'openai'
const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY })

// After — Lightning:
import OpenAI from 'aiprox-openai'
const client = new OpenAI({ apiKey: process.env.LIGHTNINGPROX_SPEND_TOKEN }) // lnpx_...

// After — Solana:
import OpenAI from 'aiprox-openai'
const client = new OpenAI({ apiKey: process.env.SOLANA_WALLET_ADDRESS }) // base58

// After — Base:
import OpenAI from 'aiprox-openai'
const client = new OpenAI({ apiKey: process.env.BASE_WALLET_ADDRESS }) // 0x...

// Everything else stays identical:
const response = await client.chat.completions.create({
  model: 'claude-opus-4-5-20251101',
  messages: [{ role: 'user', content: 'Summarize the latest AI news' }]
})
console.log(response.choices[0].message.content)

The rail is detected automatically from the apiKey format. No config needed.

What's different

AIProx routes your task through its multi-agent orchestrator. The last user message becomes the task. The orchestrator selects the best available agent and returns the result in standard OpenAI format.

Bonus: response._aiprox contains the receipt — which agents ran, what was spent, how long it took.

console.log(response._aiprox)
// {
//   agents_used: ['lightningprox'],
//   total_sats_spent: 30,
//   duration_ms: 524,
//   receipt_id: 'orch_1773...'
// }

Auth — three rails, one constructor

| apiKey format | Rail | Payment | |---------------|------|---------| | lnpx_... | Bitcoin Lightning | Sats per request | | 0x... (40+ chars) | Base/EVM | USDC on Base | | base58 (32-44 chars) | Solana | SOL or USDC |

Get a Lightning spend token: lightningprox.com Fund a Solana wallet: solanaprox.com Register: aiprox.dev

Supported models (19 across 5 providers)

| Model | Provider | |-------|----------| | claude-opus-4-5-20251101 | Anthropic | | gpt-4-turbo | OpenAI | | meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 | Together.ai | | meta-llama/Llama-3.3-70B-Instruct-Turbo | Together.ai | | mistral-large-latest | Mistral | | magistral-medium-latest | Mistral (Reasoning) | | gemini-2.5-flash | Google | | gemini-2.5-pro | Google | | (+ 11 more) | |

Full list: aiprox.dev/api/agents

Error handling

try {
  const r = await client.chat.completions.create({ ... })
} catch (err) {
  if (err.code === 'insufficient_balance') {
    // err.status === 402 — top up at aiprox.dev
  }
  if (err.code === 'no_agent_available') {
    // err.status === 404 — no agent for this capability
  }
}

Docs

aiprox.dev · Built by LPX Digital Group LLC