> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://developers.alephant.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://developers.alephant.io/_mcp/server.

# Language Models

> Configure the OpenModels catalog, choose available language models, and set a deployment default.

Language Models is where Admins choose the OpenModels catalog entries that the deployment can use. It supports a multi-model setup: make more than one approved model available and select a sensible default for ordinary requests.

## Model governance principles

More models are not always better. In production, expose only approved models with clear use cases, acceptable cost, and acceptable data boundaries.

| Decision              | Recommendation                                                                                   |
| --------------------- | ------------------------------------------------------------------------------------------------ |
| Default model         | Choose a stable, cost-controlled model that fits most chat use cases.                            |
| High-capability model | Enable only for complex reasoning, long context, or high-value Agents.                           |
| Low-cost model        | Use for summarization, classification, and lightweight Q\&A where lower cost or latency matters. |
| Self-hosted model     | Use when data boundary, network access, or cost predictability is more important.                |
| Model deprecation     | Check dependent Agents, capabilities, and workflows before disabling it.                         |

## Before you begin

Confirm that the credential required by each service is present and current. Credentials provide secure service access; this page controls which catalog models the deployment selects from that access.

## Configure catalog models

Open **Models & Capabilities**, then open **Language Models**. Review the OpenModels catalog, enable the models that fit the deployment's requirements, and associate each selection with a working credential where needed. Keep the enabled set understandable so Curators can make intentional model choices.

## Choose the default model

Set a default model for requests that do not specify one. Consider capability, quality, latency, cost, and the model's availability for the deployment. A default does not prevent approved Agents or workflows from using another enabled model when their configuration requires it.

## Verify availability

After saving, confirm that the selected models appear where authorized people configure or use them. Test a representative request, then revisit the credential and catalog selection if the model is unavailable or cannot complete the expected task.

## Related pages

* [Model Credentials](/aivis/models/model-credentials) explains how AIvis establishes secure access to model services.
* [Generation & Chat Capabilities](/aivis/models/generation-and-chat) describes capability-specific settings beyond language-model selection.