Setup Guide
fal.ai API Key on OpenClaw & Hermes Agent: AI Image & Video Generation
Add your fal.ai key on the API Keys page to give your OpenClaw or Hermes Agent instance access to the full fal model catalog — FLUX, GPT Image 2, Seedream v4, and Kling 2.1 video — all billed pay-as-you-go to your fal account. Hermes Agent specifics are in their own section below.
What Is fal.ai?
fal.ai is a hosted inference platform built for speed. It runs the most popular open-source image and video generation models on dedicated GPU infrastructure, typically returning results in seconds rather than minutes. You pay per generation with no monthly commitment, and the API follows a simple REST pattern that the OpenClaw gateway calls on your behalf (the Hermes Agent gateway does the same when you ask it to generate an image or video).
fal is particularly well-known for fast FLUX variants and for hosting cutting-edge models — like Kling 2.1 video and Seedream v4 from ByteDance — before they become widely available elsewhere. Because OpenClaw and Hermes Agent route every media request through your own fal key, your usage and billing stay entirely under your fal dashboard.
What Your fal Key Unlocks on OpenClaw and Hermes Agent
Once saved, your key enables the following models inside your OpenClaw or Hermes Agent instance’s image and video settings:
- GPT Image 2 —
fal-ai/gpt-image-2, from $0.02 per image - Nano Banana Pro / Gemini 3 Pro Image —
fal-ai/nano-banana-pro, $0.15 per image - FLUX 2 Pro —
fal-ai/flux-2-pro, $0.03 per image - Seedream v4 —
fal-ai/bytedance/seedream/v4, $0.04 per image - Flux Kontext Max —
fal-ai/flux-kontext/max, $0.06 per image — best for image editing tasks - Flux Pro 1.1 — $0.04 per image
- Qwen-Image — $0.02 per image — cheapest option in the catalog
- Flux Dev — $0.025 per image
- Kling 2.1 video —
fal-ai/kling-video/v2.1/master/text-to-video, from $1.40 per 5-second clip
All prices are fal’s published rates billed directly to your fal account. OpenClaw Launch does not mark them up.
How to Get a fal API Key
- Go to fal.ai/dashboard/keys — Open fal.ai/dashboard/keys in your browser. You will need to sign in or create a free account first.
- Create an account — Click “Sign Up” and register with your email or Google OAuth. No credit card is required to sign up, but you will need to add a payment method before generating paid content.
- Generate a new key — On the Keys page, click “Create new key”. Give it a name (for example, openclaw or hermes-agent) and click confirm.
- Copy the key immediately — fal shows the full key only once. It starts with
key_followed by a long alphanumeric string. Copy it to your clipboard now. - Add a payment method — In the fal dashboard under Billing, add a credit or debit card so your generations are not blocked by a zero balance.
Add Your fal Key on OpenClaw Launch
- Sign in at openclawlaunch.com and open the API Keys page.
- Find the fal card. Paste your
key_...key into the input field and click Save. We validate the key against fal’s API before storing it, so you will see an error immediately if something is wrong. - Open your instance’s image or video settings and pick the fal model you want as the default (for example, FLUX 2 Pro for images or Kling 2.1 for video). The change applies to your running instance without a redeploy.
Your key is encrypted at rest using AES-256-GCM and applies to all existing instances on your account — no redeploy needed. Each running instance restarts once, for a few seconds, to load the key.
Using fal.ai with Hermes Agent
The setup is the same on Hermes Agent: create the fal key, save it on the API Keys page, and it applies to all existing Hermes Agent instances on your account without a redeploy (each running bot restarts briefly to load it). The model IDs and prices above are identical. The differences are small:
- Open your Hermes Agent instance’s image or video settings and pick the fal model you want as the default (for example, FLUX 2 Pro for images or Kling 2.1 for video).
- Provider keys are independent on Hermes Agent too. You can keep an OpenRouter or MiniMax key for chat and a fal key for image and video generation active on the same instance simultaneously.
- Want cinematic video as well? See Hermes Agent + Runway to add Runway Gen-4.5 to your Hermes instance.
- For all supported provider keys on Hermes Agent, see the Hermes Agent BYOK overview.
Troubleshooting
401 Unauthorized or Invalid Key
- Confirm your key starts with
key_. fal keys do not start withsk-or any other prefix. - Make sure you copied the full key. fal shows it only once at creation — if you missed it, revoke the old key and generate a new one.
- Check for leading or trailing whitespace when pasting into the API Keys form.
Model Not Found or Unavailable
- fal occasionally deprecates or renames model paths. If a specific model ID stops working, check the fal model catalog for the current ID.
- Some fal models are gated and require your account to be approved. Check your fal dashboard for any pending access requests.
Quota or Balance Errors
- fal is pay-as-you-go. If your account balance is zero or your card is declined, generation requests will fail with a 402 or payment error.
- Check your fal Billing page and top up if needed.
- fal also enforces per-minute rate limits on lower-tier accounts. If you hit them, wait a moment and retry.
Frequently Asked Questions
Do I need a fal key to use images on OpenClaw?
A fal key is required to use fal-hosted models (FLUX, Kling, Seedream, etc.). If you have an OpenAI key saved under a separate provider card, you can use GPT Image 2 through that route instead. The fal key specifically unlocks the fal model catalog.
How much does fal image generation cost?
Prices vary by model: Qwen-Image is $0.02 per image (cheapest), FLUX 2 Pro is $0.03, Seedream v4 is $0.04, and Flux Kontext Max is $0.06 (best for editing). Kling 2.1 video starts at $1.40 per 5-second clip. All charges go directly to your fal account; OpenClaw Launch does not take a cut of media costs.
Do I need a fal key to use images on Hermes Agent?
A fal key is required to use fal-hosted models (FLUX, Kling, Seedream, etc.) on Hermes Agent as well. If you have an OpenAI key saved under a separate provider card, you can use GPT Image 2 via that route instead.
Will saving a fal key restart my running instances?
Briefly, yes. Saving or removing a key on the API Keys page pushes it to every running OpenClaw and Hermes Agent instance on your account, and each one restarts once (a few seconds) to load it. Your conversations, files and settings are kept, and there is nothing to redeploy.
Can I use fal alongside a chat provider like OpenRouter?
Yes. Provider keys are independent. You can have an OpenRouter key (or MiniMax on Hermes Agent) for chat, a fal key for image and video generation, and they will both be active on the same instance simultaneously.
What's Next?
- Hermes Agent + Runway — Add Runway Gen-4.5 cinematic video to your Hermes instance
- OpenClaw + Runway — Add Runway Gen-4.5 cinematic video to your instance
- OpenClaw BYOK overview — See all supported provider keys and how BYOK works
- Hermes Agent BYOK overview — See all supported provider keys and how BYOK works on Hermes Agent
- API Keys page — Manage all your provider keys in one place