FAQ

Conflux Router is a unified AI gateway that gives you one API endpoint for 16 models across 5 providers. Instead of managing separate accounts with OpenAI, Anthropic, Google, Cerebras, and Groq, you get a single OpenAI-compatible API with credit-based billing and smart routing.

Sign up at theconflux.com, grab your API key from the dashboard, and start making requests to /v1/chat/completions. You get 500 free credits to try it out — no credit card required.

We're fully OpenAI-compatible. If you've used the OpenAI SDK or any OpenAI-compatible client, you already know how to use Conflux Router. Just swap the base URL to https://theconflux.com/v1 and use your Conflux API key.

We route to 16 models across 5 providers. See the Models page for the full list with pricing and capabilities.

Credits are Conflux Router's universal currency. Every API call costs credits based on the model used and the number of tokens processed. One credit = $0.01 in value. Credits abstract away provider-specific pricing so you deal with one number, not five different rate cards.

Free tier gets 500 credits. Power plan ($24.99/mo) includes 10,000 credits with overage at $0.003/credit. Pro plan ($49.99/mo) includes 30,000 credits with overage at $0.002/credit.

Credits do not roll over between billing periods. Your monthly allocation resets on your billing date.

On the Free tier, requests return 402 Payment Required once credits are exhausted. On Power and Pro plans, overage billing kicks in automatically — you're charged per extra credit at your plan's overage rate.

Credit packs for one-time purchases are coming soon. Currently, credits are allocated through subscription plans.

Live keys (cf_live_...) make real API calls and are billed normally. Test keys (cf_test_...) are for development — they validate your setup and return mock responses without consuming credits.

Yes. When generating a key, you can set an optional expiration timestamp. Expired keys are automatically rejected. You can also revoke any key instantly from the dashboard or the /v1/keys endpoint.

Yes. Generate as many keys as you need — one per project, per environment, per team member, whatever works for you. Each key is tracked independently in usage logs.

Keys are hashed with SHA-256 before storage. We never store your raw key. If you lose a key, you'll need to generate a new one — we can't recover the original.

Rate limits are enforced per API key at 60 requests per minute. If you exceed the limit, you'll receive a 429 Too Many Requests response. Limits may vary by plan tier in the future.

Use the GET /v1/usage endpoint to retrieve your request history, token counts, and credit consumption. The dashboard also shows real-time usage graphs.

Use the GET /v1/credits endpoint to see your current balance, plan details, and usage breakdown.

When you request a model, Conflux Router checks provider health (monitored every 60 seconds) and routes to the best available provider. If the primary provider is down, we automatically failover to the next healthy provider in the chain.

Primary model → fallback model → next provider → 503 Service Unavailable. We try every reasonable path before giving up.

Not currently. Conflux Router manages provider selection automatically based on health, availability, and cost optimization. Provider pinning is on the roadmap.

If every provider for a given model is unhealthy, the API returns a 503 Service Unavailable. This is rare — our health monitoring and failover chain are designed to prevent it.

Conflux Router does not store your prompts or model responses. Requests are forwarded directly to providers. We log metadata (token counts, credit costs, latency) for billing and monitoring, but not content.

Payment processing is handled by Stripe. We never store full card numbers. Credit balances and transaction logs are stored in our database with standard access controls.

Yes. Contact support to initiate account deletion. All API keys are revoked, and your data is purged within 30 days.