Using Fireworks AI with OpenClaw

Prerequisites

• ✓An existing OpenClaw installation with the CLI available in your shell.
• ✓A Fireworks account with an active API key.
• ✓Access to the shell or environment where your OpenClaw Gateway or agents run, so you can set FIREWORKS_API_KEY for that process.

Steps

1. 1

   Understand how Fireworks plugs into OpenClaw

   Before you start wiring config, align on how OpenClaw talks to Fireworks. ai/inference/v1`.

   5 Turbo router.

   textCopy

   Provider: `fireworks`
   Auth: `FIREWORKS_API_KEY`
   API: OpenAI-compatible chat/completions
   Base URL: `https://api.fireworks.ai/inference/v1`
   Default model: `fireworks/accounts/fireworks/routers/kimi-k2p5-turbo`

2. 2

   Onboard Fireworks interactively with the OpenClaw CLI

   Use the OpenClaw onboarding flow to register your Fireworks API key once and let OpenClaw store it in its config. This command prompts you for the key and sets the Fire Pass starter model as the default, so your agents can call Fireworks without extra wiring.

   Run it from a shell where you can paste your Fireworks key.

   bashCopy

   openclaw onboard --auth-choice fireworks-api-key

3. 3

   Automate Fireworks setup with the non-interactive onboarding command

   If you script deployments or CI, use the non-interactive onboarding variant. It reads the Fireworks key from the FIREWORKS_API_KEY environment variable, configures OpenClaw in local mode, skips health checks, and accepts the risk flags automatically.

   This is the command you drop into provisioning scripts so Fireworks is ready without prompts.

   bashCopy

   openclaw onboard --non-interactive \
     --mode local \
     --auth-choice fireworks-api-key \
     --fireworks-api-key "$FIREWORKS_API_KEY" \
     --skip-health \
     --accept-risk

4. 4

   Make FIREWORKS_API_KEY available to your Gateway environment

   If your OpenClaw Gateway runs under systemd, launchd, or another supervisor, the environment from your interactive shell does not automatically carry over. profile`.

   Configure the service environment or secrets store so the key is present whenever the Gateway starts.

   textCopy

   If the Gateway runs outside your interactive shell, make sure `FIREWORKS_API_KEY` is available to that process too. A key sitting only in `~/.profile` will not help a launchd/systemd daemon unless that environment is imported there as well.

5. 5

   Use the built-in Fireworks Kimi router as your default model

   5 Turbo router on Fireworks. You can reference this model id directly in your agent defaults to standardize on a high-context, text-and-image capable model.

   This is the same model the onboarding flow sets as the default starter model.

   textCopy

   fireworks/accounts/fireworks/routers/kimi-k2p5-turbo

6. 6

   Configure a custom Fireworks model id in your agent defaults

   When Fireworks publishes new models or routers, you can point OpenClaw at them without waiting for a catalog update. Prefix the exact Fireworks model or router id with `fireworks/` and set it as your agent’s primary model.

   This JSON snippet shows how to wire the bundled Kimi router, and you can swap the id for any other Fireworks model.

   jsonCopy

   {
     agents: {
       defaults: {
         model: {
           primary: "fireworks/accounts/fireworks/routers/kimi-k2p5-turbo",
         },
       },
     },
   }

Configuration

Troubleshooting

If the Gateway runs outside your interactive shell, make sure `FIREWORKS_API_KEY` is available to that process too. A key sitting only in `~/.profile` will not help a launchd/systemd daemon unless that environment is imported there as well.

OpenClaw accepts dynamic Fireworks model ids too. Use the exact model or router id shown by Fireworks and prefix it with `fireworks/`.

Frequently asked questions