Skip to Content
Haven Documentation is the shared home for product guides, plugin setup, and release docs.
PluginsAIO AgentFree LLM

Free LLM

The easiest free LLM option for AIO Agent is Google Gemini through Google AI Studio . Google provides a Gemini API free tier for developers and small projects, which is enough for testing AIO Agent, setting up support flows, and trying owner/admin chat before deciding whether you need paid quota.

Free tier availability, model access, and rate limits are controlled by Google and can change. If you are setting this up for a customer, confirm the current limits on Google’s Gemini API pricing and rate-limit pages.

Get A Free Gemini API Key

  1. Go to Google AI Studio .
  2. Sign in with the Google account you want to use for the API key.
  3. Open the API key page, or use Google’s Get a Gemini API key  shortcut.
  4. Click Create API key.
  5. Select an existing project, or let Google AI Studio create/import one.
  6. Copy the generated key and keep it private.

Google’s API key docs explain that new users may get a default Google Cloud project and API key after accepting the terms. Existing Google Cloud users may need to import or select a project first.

Configure AIO Agent

In the Athena dashboard, open Configuration -> AIO Agent and set:

{
  config: {
    enabled: true,
    llm_provider: "gemini",
    llm_api_key: "PASTE_YOUR_GEMINI_KEY_HERE",
    llm_model: "gemini-2.5-flash"
  }
}

Save the dashboard config and restart AthenaBot.

The llm_model field is free text. If Google changes model names or gives you access to a newer Gemini model, paste the exact model name from Google AI Studio instead of waiting for the dashboard dropdown to include it.

For a safer production setup, leave llm_api_key blank in the dashboard and set the key as an environment variable on the machine running AthenaBot:

AIO_AGENT_GEMINI_API_KEY=your_key_here

You can also use:

AIO_AGENT_LLM_API_KEY=your_key_here

AIO_AGENT_LLM_API_KEY works for any selected provider. The Gemini-specific variable is useful when the same server runs multiple provider configs.

Use:

llm_provider: "gemini",
llm_model: "gemini-2.5-flash"

gemini-2.5-flash is the recommended default for AIO Agent because it is fast and cost-friendly for ticket replies, owner questions, knowledgebase answers, and normal support automation.

Important Free-Tier Notes

  • Free tier is best for testing, demos, small communities, and early customer setup.
  • Free tier can have lower rate limits than paid usage.
  • Google may use free-tier API input and output to improve its products.
  • Do not enable billing unless you want paid quota.
  • If the bot starts seeing 429 rate-limit errors or temporary 503 high-demand errors, either wait and retry later or move the key/project to a paid tier.

Keep The Key Safe

  • Never paste the real key in screenshots, public support channels, or docs.
  • Never commit the key to GitHub or any plugin package.
  • Keep the key server-side. AIO Agent runs inside AthenaBot, so the key should stay in the dashboard config, Mongo-backed config, or server environment.
  • In Google Cloud Console, restrict the key to the Generative Language API when possible.
  • Rotate the key if it is ever leaked.

Quick Test

After saving the key and restarting AthenaBot, mention the bot from an allowed owner/admin account:

@Athena Dev where can I find the docs?

Then test a normal AIO Agent request:

@Athena Dev summarize the last 10 messages here

If the bot replies, the Gemini key is working.

Optional Gemini Embeddings

If a customer’s knowledgebase is large, you can keep Gemini as the main LLM and also use Gemini embeddings for hybrid retrieval:

knowledgebase_mode: "hybrid",
knowledgebase_embeddings_enabled: true,
knowledgebase_embedding_provider: "llm",
knowledgebase_embedding_model: "text-embedding-004"

Leave embeddings disabled for small knowledgebases. Normal search mode is cheaper and simpler because it does not need extra embedding calls.

SetupConfiguration