Hermes

Hermes is Nous Research's terminal-based coding agent. It ships an AI Gateway provider, so you can reach every model in the gateway catalog from Hermes with one API key and track spend in one place.

1. Create an API key

   Go to the AI Gateway section in the Vercel dashboard sidebar and click API keys to create a new API key.

2. Add the key to Hermes

   Hermes reads AI_GATEWAY_API_KEY from your environment and from ~/.hermes/.env. Export it in your shell:

   Terminal

   export AI_GATEWAY_API_KEY="your-ai-gateway-api-key"

   To persist the key across sessions, add the same line to ~/.hermes/.env instead.

3. Select AI Gateway as your provider

   Run the model picker:

   Terminal

   hermes model

   Choose Vercel AI Gateway, then pick a model. Hermes fetches the live catalog with current pricing from the gateway, filtered to language models that support tool calling. Image, video, embedding, and reranking models from the models page don't appear in the picker.

4. Start Hermes

   You can also skip the picker and name the provider and model directly:

   Terminal

   hermes --provider ai-gateway -m openai/gpt-5.6-sol

   The provider ID is ai-gateway. Hermes also accepts vercel and vercel-ai-gateway as aliases. Model IDs use the gateway's creator/model-name format.

5. (Optional) Set a fallback provider

   Hermes supports ai-gateway in fallback chains. Add it to fallback_model in ~/.hermes/config.yaml, or run hermes fallback to configure it interactively. When a fallback activates, Hermes swaps the model and provider mid-session without losing your conversation.

6. (Optional) Monitor usage and spend

   View your usage, spend, and request activity in the AI Gateway section in the Vercel dashboard sidebar. See the observability documentation for more details.

   Hermes identifies itself to the gateway, so its requests are attributable in your analytics.

• Browse the models catalog for available models
• Learn about model routing and fallbacks
• Configure spend limits on your API key