> ## Documentation Index
> Fetch the complete documentation index at: https://docs.deepinfra.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Hermes Agent

> Use DeepInfra models with Hermes Agent by adding a custom OpenAI-compatible provider.

<Note>
  Don't want to host it yourself? DeepInfra can run Hermes for you — see [Hosted Agents: Hermes-Agent](/agents/hermes-agent).
</Note>

[Hermes Agent](https://github.com/NousResearch/hermes-agent) is a self-hosted, self-improving autonomous agent by Nous Research. DeepInfra isn't a built-in provider, but Hermes works with any OpenAI-compatible endpoint (`/v1/chat/completions`) — so you add DeepInfra as a `custom` provider and use any [LLM from our catalog](https://deepinfra.com/models/text-generation).

## Configure `~/.hermes/config.yaml`

```yaml theme={null}
model:
  default: deepseek-ai/DeepSeek-V4-Flash
  provider: custom
  base_url: https://api.deepinfra.com/v1/openai
  api_key: ${DEEPINFRA_TOKEN}
  context_length: 1048576
```

Get your API key from the [Dashboard](https://deepinfra.com/dash/api_keys).

<Note>
  * `provider: custom` is required for any non-built-in OpenAI-compatible endpoint, and `base_url` overrides `provider`.
  * The DeepInfra model id passes through verbatim in `default` — no reformatting needed.
  * `api_key` falls back to the `OPENAI_API_KEY` environment variable if omitted, and secrets can live in `~/.hermes/.env` instead of the YAML.
  * Set `context_length` explicitly. Above it's the model's full 1M-token window; Hermes needs roughly 64k minimum for agent functionality. Check a model's window via [`/v1/openai/models?filter=with_meta`](https://api.deepinfra.com/v1/openai/models?filter=with_meta\&sort_by=openclaw).
</Note>

## Interactive alternative

Instead of editing the YAML by hand, run the model wizard and pick the custom endpoint:

```bash theme={null}
hermes model
# Choose: "Custom endpoint"
# Base URL: https://api.deepinfra.com/v1/openai
# API key:  <your DeepInfra API token>
# Model:    deepseek-ai/DeepSeek-V4-Flash
```

Hermes persists the choice and reuses it on every run.

## Run it

```bash theme={null}
hermes
```

## Learn more

<CardGroup cols={2}>
  <Card title="AI Providers" icon="plug" href="https://hermes-agent.nousresearch.com/docs/integrations/providers">
    Hermes' provider overview and custom-endpoint option.
  </Card>

  <Card title="Configuration" icon="gear" href="https://hermes-agent.nousresearch.com/docs/user-guide/configuration">
    The full `config.yaml` reference.
  </Card>

  <Card title="Chat Completions" icon="comments" href="/chat/overview">
    DeepInfra's OpenAI-compatible API.
  </Card>

  <Card title="Authentication" icon="key" href="/account/authentication">
    API keys and scoped JWTs.
  </Card>
</CardGroup>