theopenco

One API for every model — and the products we build on top of it

Website · Models · Providers · Pricing · Docs · Status

We build open-source infrastructure for the multi-model era. One OpenAI-compatible endpoint routes to every major provider with automatic fallback, caching, cost analytics and guardrails — and everything else we ship runs on it.

flowchart LR
A["DevPass Code<br/>terminal agent"] --> G
B["Lounge<br/>chat &amp; studios"] --> G
C["Your app<br/>OpenAI SDK"] --> G
G["<b>LLM Gateway</b><br/>routing · fallback · caching · analytics"] --> P[("every major<br/>model &amp; provider")]

One endpoint in front of every model. Point the OpenAI SDK at us, change the base URL, and you're done — no rewrites, no per-provider glue code. Requests retry transparently on a healthy provider when one degrades, and every token lands in your usage dashboard.

• Drop-in compatible — OpenAI-format chat completions, streaming, tool calls, JSON mode, vision, reasoning
• Automatic fallback — a degraded provider retries on a healthy one, mid-request
• Cost-aware analytics — spend, tokens, cache hits and errors broken down by model, provider and project
• Pay at provider rates — flat 5% platform fee on credits, or free with your own keys (BYOK)
• Guardrails & audit logs — content filtering, compliance-aware routing, org-wide audit trail
• Self-host anything — the core is AGPL-3.0 and ships as a single container

curl https://api.llmgateway.io/v1/chat/completions \
-H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "model": "gpt-4o", "messages": [{ "role": "user", "content": "Hello!" }] }'

docker run -d --name llmgateway \
-p 3002:3002 -p 4001:4001 \
-e AUTH_SECRET="$(openssl rand -base64 32 | tr -d '\n')" \
-e GATEWAY_API_KEY_HASH_SECRET="$(openssl rand -base64 32 | tr -d '\n')" \
-v llmgateway_postgres:/var/lib/postgresql/data \
-v llmgateway_redis:/var/lib/redis \
ghcr.io/theopenco/llmgateway-unified:latest

One AI coding subscription. Every model. Three flat prices. Every dollar becomes $3 of model usage at provider rates — metered transparently, no token math, no lock-in. Works in DevPass Code (our first-party terminal agent) and drops into every OpenAI-compatible tool you already run.

npm i -g devpass-code # or: brew install theopenco/tap/devpass-code
devpass-code auth login # opens your browser — no keys to copy, models built in

→ devpass.llmgateway.io · DevPass Code repo · setup guide

Every frontier model. One membership. Chat with the models you already pay three subscriptions for, run the same prompt across several at once in group chat, and generate images, video and audio from the same place — with knowledge-base projects, canvas and voice calls built in.

→ lounge.llmgateway.io

The core platform is AGPL-3.0. Enterprise features under ee/ are commercially licensed — get in touch.