← Home

Setup Guide

Add a Groq API Key to Hermes Agent

Groq's LPU hardware makes your Hermes Agent replies feel instant — typically under 500ms to the first token on Llama 3.3 70B. Paste your gsk_ key on the API Keys page and all your Hermes Agent instances switch to Groq straight away, no container restart needed.

What Is Groq?

Groq is an AI inference company that designed the LPU (Language Processing Unit), a purpose-built chip that runs transformer models dramatically faster than GPU-based alternatives. Real-world first-token latency for 70B models sits under 500 ms on Groq, and sustained generation speed often exceeds several hundred tokens per second — noticeably faster than any GPU cloud for the same model class.

Groq includes a free tier with generous daily rate limits, which is more than enough for most Hermes Agent deployments. If your agent handles heavy traffic, pay-as-you-go billing unlocks higher caps. All models below are accessible on the free tier subject to those limits.

What Your Groq Key Unlocks in Hermes Agent

After saving a Groq key on the API Keys page, the following models are available as the active model for your Hermes Agent instances. Every model supports a 131K context window.

  • Llama 3.3 70B Versatile (llama-3.3-70b-versatile) — the default Groq model; strong general-purpose reasoning, instruction following, and code generation
  • Llama 3.1 8B Instant (llama-3.1-8b-instant) — smallest footprint; best for simple Q&A, high-volume tasks, or cost-sensitive workloads
  • Kimi K2 (moonshotai/kimi-k2-instruct) — Moonshot AI's model optimised for deep reasoning and coding tasks
  • GPT-OSS 120B (openai/gpt-oss-120b) — OpenAI's open-weights 120B model served at Groq speed
  • Qwen 3 32B (qwen/qwen3-32b) — Alibaba's multilingual and math-capable model with strong CJK support

How to Get a Groq API Key

  1. Open the Groq console — Go to console.groq.com/keys. No account yet? Click Sign Up — it takes under a minute and no credit card is required.
  2. Create a new API key — Click Create API Key, give it a recognisable name such as “Hermes Agent — OpenClaw”, and confirm.
  3. Copy the key immediately — Groq only shows the full key once at creation time. Copy it before navigating away.
  4. Note the prefix — Your key begins with gsk_. OpenClaw uses this prefix to validate the key before saving it.

Add Your Groq Key to Hermes Agent on OpenClaw Launch

  1. Sign in at openclawlaunch.com and navigate to the API Keys page.
  2. Find the Groq card, paste your gsk_ key, and click Save. OpenClaw validates the key live against the Groq API before encrypting and storing it.
  3. After saving, click “Set as primary model” and choose your preferred Groq model. The change pushes to all running Hermes Agent instances without a container restart.
  4. Your key is stored encrypted at rest. OpenClaw never logs or exposes it in plaintext.

Running OpenClaw instances instead? See the Groq for OpenClaw guide for the OpenClaw-specific walkthrough.

Troubleshooting

401 Unauthorized — Invalid API Key

  • Confirm the key starts with gsk_. Partial copies or organisation tokens from the Groq dashboard will not authenticate.
  • Check for invisible whitespace. Paste the key into a plain-text editor first to strip any stray characters before entering it on the API Keys page.
  • If you revoked the key and created a new one, the old value is permanently invalid. Update it on the API Keys page.

Model Not Found or Not Available

  • Use the exact model IDs shown above (e.g. llama-3.3-70b-versatile). Groq model IDs are case-sensitive and must match exactly.
  • Groq periodically rotates preview models. If a non-default model stops working, fall back to llama-3.3-70b-versatile while you investigate.

Rate Limit or Daily Quota Exceeded

  • Free-tier keys have per-minute and daily token limits. Enable billing in the Groq console to lift the daily cap.
  • Switching from Llama 3.3 70B to Llama 3.1 8B Instant lowers token consumption significantly and extends the free allowance.
  • OpenClaw surfaces 429 errors in Hermes Agent instance logs — check there if your agent stops responding.

Frequently Asked Questions

Is Groq free to use with Hermes Agent?

Yes — Groq has a free tier with daily token limits suitable for most personal and prototype workloads. Sign up at console.groq.com/keys without a credit card. Pay-as-you-go billing is available if you need higher limits.

What does a Groq API key look like?

Groq API keys always start with gsk_ followed by a long alphanumeric string. If your key doesn't match this pattern, you may have copied an organisation ID or another value from the console rather than the API key itself.

Will adding a Groq key restart my Hermes Agent containers?

No. Saving the key on the API Keys page and clicking “Set as primary model” hot-applies the change to running Hermes Agent instances without a container restart or any downtime.

Can I switch back to another provider later?

Yes. Keys for all your providers are stored on the API Keys page. You can change the primary model at any time — the change applies instantly to running instances.

What's Next?

Instant Replies for Your Hermes Agent

Add your Groq key and your Hermes Agent starts responding in under 500ms — free tier included, no container restart needed.

Deploy with OpenClaw Launch