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

# Models and reasoning

> Discover account-supported models and configure Agent reasoning behavior.

Always call `GET /api/v2/models/supported?targetAccountId=...` before selecting a model. The response provides exact model identifiers, providers, reasoning support, and Fast-mode compatibility for the selected account.

## Agent fields

* `agentModel` is the primary model for conversations and Computer Agent work.
* With `agentS2SEngine:"openai-live"`, `agentModel` is the delegation model used for reasoning and
  tool decisions while GPT Live handles real-time audio.
* `agentThinking` enables a supported reasoning budget on a reasoning-capable primary model. Use `null` to disable it.
* `agentThinkingModel` selects the model used by Pause & Think and optional post-conversation reasoning.
* `postConvoForceThinking` uses the thinking model for summaries, reporting, and similar post-conversation work when the primary model is not already reasoning-enabled.

## Fast processing

`agentFastMode` is optional and disabled by default. Enable it only when the selected model reports
`supportsFastMode:true`. Vida then requests OpenAI Fast processing for compatible model calls,
including GPT Live delegation. Billing follows the service tier OpenAI actually returns, so confirm
the account's product plan before enabling it for customer traffic.

Model availability can vary by account and compliance requirements. Store the exact identifier returned by discovery; do not copy a static model list into an integration.

After updating staging, read it again and confirm the selected model and reasoning settings. Test latency and behavior before publishing.