Skip to main content

OpenAI Setup

OpenAI is available in AxonFlow Community and Enterprise. It is the simplest cloud provider to start with for Proxy Mode, MAP, and Gateway Mode examples.

Runtime Defaults​

AxonFlow's OpenAI provider defaults to:

  • Provider name: openai
  • Default model: gpt-4o
  • Default endpoint: https://api.openai.com

If you want a different model, set it explicitly in environment variables or YAML.

Environment Variables​

export OPENAI_API_KEY=sk-your-api-key
export OPENAI_MODEL=gpt-4o
export OPENAI_TIMEOUT_SECONDS=120    # Request timeout in seconds (default: 120)
export OPENAI_ENDPOINT=https://api.openai.com  # Custom endpoint for proxies or compatible APIs

YAML Configuration​

version: "1.0"

llm_providers:
  openai:
    enabled: true
    credentials:
      api_key: ${OPENAI_API_KEY}
    config:
      model: ${OPENAI_MODEL:-gpt-4o}

Good Fits​

  • General-purpose chat and agent tasks
  • Strong default for early community trials
  • Teams that want broad SDK support across TypeScript, Python, Go, and Java
  • Internal copilots, customer support agents, and application backends that need OpenAI governance without rewriting model-call logic

Proxy Mode​

TypeScript​

import { AxonFlow } from '@axonflow/sdk';

const axonflow = new AxonFlow({
  endpoint: 'http://localhost:8080',
  clientId: process.env.AXONFLOW_CLIENT_ID,
  clientSecret: process.env.AXONFLOW_CLIENT_SECRET,
});

const response = await axonflow.proxyLLMCall({
  userToken: 'user-123',
  query: 'Summarize the main risks in this API design.',
  requestType: 'chat',
  context: {
    provider: 'openai',
    model: 'gpt-4o',
  },
});

console.log(response.data);

Python​

from axonflow import AxonFlow

async with AxonFlow(
    endpoint="http://localhost:8080",
    client_id="demo-client",
    client_secret="demo-secret",
) as client:
    response = await client.proxy_llm_call(
        user_token="user-123",
        query="Summarize the main risks in this API design.",
        request_type="chat",
        context={"provider": "openai", "model": "gpt-4o"},
    )
    print(response.data)

Gateway Mode​

Use Gateway Mode when you want AxonFlow to approve and audit the request but your application to call OpenAI directly.

import { AxonFlow } from '@axonflow/sdk';
import OpenAI from 'openai';

const prompt = 'Summarize the main risks in this API design.';

const axonflow = new AxonFlow({
  endpoint: 'http://localhost:8080',
  clientId: process.env.AXONFLOW_CLIENT_ID,
  clientSecret: process.env.AXONFLOW_CLIENT_SECRET,
});

const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

const ctx = await axonflow.getPolicyApprovedContext({
  userToken: 'user-123',
  query: prompt,
});

if (!ctx.approved) {
  throw new Error(`Blocked: ${ctx.blockReason}`);
}

const approvedPrompt =
  typeof ctx.approvedData.query === 'string' ? String(ctx.approvedData.query) : prompt;

const completion = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: approvedPrompt }],
});

const output = completion.choices[0].message.content ?? '';

await axonflow.auditLLMCall({
  contextId: ctx.contextId,
  responseSummary: output.slice(0, 200),
  provider: 'openai',
  model: 'gpt-4o',
  tokenUsage: {
    promptTokens: completion.usage?.prompt_tokens ?? 0,
    completionTokens: completion.usage?.completion_tokens ?? 0,
    totalTokens: completion.usage?.total_tokens ?? 0,
  },
  latencyMs: 250,
});

Notes for Production Teams​

  • AxonFlow uses the configured OpenAI provider for Proxy Mode and MAP.
  • Request-level context.provider = "openai" is a routing hint unless you also set context.strict_provider = true.
  • If your OpenAI account has limited model access, keep OPENAI_MODEL or config.model aligned with what the account can actually use.

Troubleshooting​

IssueCauseFix
401 UnauthorizedInvalid or expired API keyVerify OPENAI_API_KEY
429 Rate LimitedToo many requestsReduce concurrency or upgrade OpenAI plan
Model not foundModel ID not accessibleCheck model access in OpenAI dashboard

See Also​

Provider Rollout Checklist​

Before this provider path becomes production traffic, connect it to the runtime operating model:

Last updated on