AI Configuration

Overview

The AI Configuration page lets you choose which AI model powers your test executions and test generation, manage your own API keys (Bring Your Own Key), and monitor token usage and costs. For VibeView-provided runs the cost shown is the usage credit actually deducted from your balance; for BYOK runs VibeView deducts nothing, so the cost is shown as an estimate (billed directly by your provider). See Usage Credits.

Features

Default Model

Select the AI model used for all AI-powered features: test execution, hybrid replay fallback, and test generation from recordings. The dropdown shows only vision-capable models that you have access to (through your own API keys or VibeView-hosted keys). Changing the selection takes effect immediately for all future runs.

API Keys (BYOK)

Bring Your Own Key lets you connect your existing provider accounts so that AI requests are billed directly to your provider instead of through VibeView.

Supported providers:

• Anthropic (Claude) — Claude Sonnet 4.6, Claude Opus 4.6, Claude Haiku 4.5
• OpenAI (GPT) — GPT-5, GPT-5 Mini, GPT-4o, o4-mini
• Google (Gemini) — Gemini 2.5 Flash, Gemini 2.5 Pro, Gemini 2.0 Flash
• OpenRouter — Access models from all major providers through a single API key. Includes Claude, GPT, Gemini, DeepSeek R1, Llama 4, and more. One key instead of managing multiple provider accounts.

Each stored key shows the provider badge and a masked key hint (e.g., sk-...abc). Keys are encrypted at rest.

Usage Tracking

The Usage section provides a breakdown of AI consumption:

• Summary cards — Total runs, input tokens, output tokens, and total cost, shown in USD. The total cost combines VibeView-provided (actual, deducted) and BYOK (estimated) cost; the two are broken out in the split below.
• VibeView vs BYOK split — Two panels comparing runs, tokens, and cost. The VibeView cost is the credit actually deducted for provided runs; the BYOK cost is an estimate at public list pricing — VibeView deducts nothing for BYOK runs, and your provider bills you directly, so the exact amount may differ.
• Per-model breakdown — A table listing each model with its run count and cost (the credit deducted per model).

A time-range selector (7d / 30d / 90d / All) at the top of the Usage section controls the period all of these figures cover.

Token counts are displayed in human-readable format (e.g., 1.2M, 45.3K).

How to Add an API Key

1. Navigate to Settings > AI Configuration.
2. In the API Keys card, click Add Key.
3. Select a Provider from the dropdown (Anthropic, OpenAI, Google, or OpenRouter).
4. Paste your API key into the API Key field.
5. Click Save. A success toast confirms the key was stored.

The new key unlocks additional models from that provider in the Default Model dropdown.

Tip: If you want access to multiple providers, OpenRouter is the easiest option — one key gives you access to Claude, GPT, Gemini, and many more models.

How to Remove an API Key

1. In the API Keys list, click the trash icon next to the key you want to delete.
2. A confirmation dialog appears warning that BYOK models for that provider will become unavailable.
3. Click Delete to confirm.

How to Change the Default Model

1. In the Default Model card, open the model dropdown.
2. Select the desired model. The dropdown lists VibeView-hosted models (billed to your usage credit) plus any models unlocked by your own API keys.
3. The change is saved automatically.

The selected model is used for:

• AI test execution (pure AI and hybrid fallback)
• Test generation from recordings

Permissions