Get started in 2 minutes
Prerequisites
- An OpenClaw installation (install guide)
- A Together AI API key — grab one at api.together.ai
Step 1: Onboard with Together AI

TOGETHER_API_KEY and store it securely for the Gateway.

Step 2: Set your default model
Using the onboard command and “QuickStart” mode you will get the default model selected by default as Kimi K2.5. Otherwise you can also change this within your OpenClaw config, setting your default model. Remember to prefix the model name with “together/“:Step 3: Launch and chat
Start the Gateway and begin chatting — via the web UI, CLI, Telegram, or WhatsApp:Environment note
If the Gateway runs as a daemon (launchd / systemd), make sureTOGETHER_API_KEY is available to that process — for example, in ~/.openclaw/.env or via env.shellEnv.
Why Together AI + OpenClaw?
Together AI gives you access to the best open-source models with high throughput and low latency. For token-hungry agentic workflows like OpenClaw, this translates to massive savings without sacrificing quality:- Kimi K2.5 — 256K context, state-of-the-art reasoning model
- DeepSeek V3.1 / R1 — top-tier coding and reasoning model
- GLM 4.7 — strong & fast all-rounder model