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

# Model Configuration

> Choose the LLM provider and model your automations run on

Point Optexity at any LLM — Gemini, Claude, GPT, or a self-hosted endpoint — by setting one environment variable. Every LLM call routes through [LiteLLM](https://docs.litellm.ai/docs/providers), so any model LiteLLM supports works without code changes.

## Environment Variables

| Variable                     | Type          | Default                        | Description                            |
| ---------------------------- | ------------- | ------------------------------ | -------------------------------------- |
| `LLM_MODEL`                  | `str`         | `gemini/gemini-3.5-flash-lite` | Primary model for every LLM call       |
| `LLM_MODEL_API_KEY`          | `str \| None` | `None`                         | API key for `LLM_MODEL`                |
| `LLM_MODEL_FALLBACK`         | `str \| None` | `None`                         | Model used when the primary call fails |
| `LLM_MODEL_FALLBACK_API_KEY` | `str \| None` | `None`                         | API key for `LLM_MODEL_FALLBACK`       |

```bash theme={null}
LLM_MODEL=anthropic/claude-sonnet-4-6
LLM_MODEL_API_KEY=YOUR_ANTHROPIC_API_KEY

LLM_MODEL_FALLBACK=openai/gpt-4.1-mini
LLM_MODEL_FALLBACK_API_KEY=YOUR_OPENAI_API_KEY
```

The primary and fallback models can live on different providers, each with its own key. Omit a key and LiteLLM reads the provider's own environment variable instead (`GEMINI_API_KEY` / `GOOGLE_API_KEY`, `ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, ...).

## Model Strings

Write models as `provider/model`:

| Model String                                      | Provider                |
| ------------------------------------------------- | ----------------------- |
| `gemini/gemini-3.5-flash-lite`                    | Google Gemini (default) |
| `gemini/gemini-2.5-flash`                         | Google Gemini           |
| `gemini/gemini-2.5-pro`                           | Google Gemini           |
| `anthropic/claude-sonnet-4-6`                     | Anthropic               |
| `anthropic/claude-haiku-4-5-20251001`             | Anthropic               |
| `openai/gpt-4.1-mini`                             | OpenAI                  |
| `bedrock/anthropic.claude-sonnet-4-20250514-v1:0` | AWS Bedrock             |

See the [LiteLLM provider list](https://docs.litellm.ai/docs/providers) for the full set of supported prefixes.

<Tip>
  Always include the provider prefix. A bare name like `gemini-2.5-flash` is passed to LiteLLM as-is and may resolve to a different provider than you expect.
</Tip>

## Resolution Order

The model for any given LLM call is resolved from the most specific setting available:

| Priority | Source                           | Example                                |
| -------- | -------------------------------- | -------------------------------------- |
| 1        | Action-level `llm_model_name`    | `extraction_action.llm.llm_model_name` |
| 2        | Task-level `llm_model_name`      | Top-level field on the automation      |
| 3        | `LLM_MODEL` environment variable | `gemini/gemini-3.5-flash-lite`         |

### Task-Level Override

Set the model once for the whole automation:

```json theme={null}
{
  "llm_model_name": "anthropic/claude-sonnet-4-6",
  "nodes": []
}
```

### Action-Level Override

Give a single action a stronger (or cheaper) model than the rest of the automation:

```json theme={null}
{
  "type": "action_node",
  "extraction_action": {
    "llm": {
      "extraction_format": { "invoice_total": "str" },
      "extraction_instructions": "Read the invoice total from the summary table.",
      "llm_model_name": "gemini/gemini-2.5-pro"
    }
  }
}
```

Actions that accept `llm_model_name`: `extraction_action.llm`, `extraction_action.pdf`, `extraction_action.locator`, `misc_action.llm_query`, and `interaction_action.captcha`.

<Info>
  Captcha solving is the one exception to the resolution order — `interaction_action.captcha` defaults to `gemini/gemini-2.5-pro` rather than falling through to `LLM_MODEL`, because captcha grids need a stronger vision model. Override `llm_model_name` on the action to change it.
</Info>

## Fallbacks

When `LLM_MODEL_FALLBACK` is set, a failed primary call is retried once on the fallback model before the action errors. Each model uses its own key, so the fallback can be on a completely different provider — useful for surviving a single provider's rate limits or outages.

```bash theme={null}
LLM_MODEL=gemini/gemini-3.5-flash-lite
LLM_MODEL_API_KEY=YOUR_GEMINI_API_KEY

LLM_MODEL_FALLBACK=anthropic/claude-haiku-4-5-20251001
LLM_MODEL_FALLBACK_API_KEY=YOUR_ANTHROPIC_API_KEY
```

Fallbacks apply to the environment-level models only. A per-task or per-action `llm_model_name` overrides the primary model; the same `LLM_MODEL_FALLBACK` still backs it up.

## Cost Tracking

Token usage and cost are reported per task from LiteLLM's pricing data. Reasoning and tool-use tokens are already counted inside completion tokens, so they are reported but never billed twice. If a model has no pricing entry in LiteLLM, tokens are still tracked and cost is reported as `0`.

## Migrating from `llm_provider`

`llm_provider` is deprecated. Existing automations that set it keep working — the provider and model are joined into a LiteLLM string — but new automations should use a single prefixed `llm_model_name`.

| Before                                                               | After                                             |
| -------------------------------------------------------------------- | ------------------------------------------------- |
| `"llm_provider": "gemini", "llm_model_name": "gemini-2.5-flash"`     | `"llm_model_name": "gemini/gemini-2.5-flash"`     |
| `"llm_provider": "anthropic", "llm_model_name": "claude-sonnet-4-6"` | `"llm_model_name": "anthropic/claude-sonnet-4-6"` |
| `"llm_provider": "openai", "llm_model_name": "gpt-4.1-mini"`         | `"llm_model_name": "openai/gpt-4.1-mini"`         |

The provider is no longer restricted to `gemini`, `anthropic`, or `openai` — any LiteLLM prefix is accepted.