Nexo Whisper

Configure AI enhancement

AI enhancement takes the raw Whisper transcription and passes it through an LLM to clean it up, format it, or transform it according to the prompt you chose.

It's completely optional — Nexo Whisper works great without it.

BYOK: you use your own API key

Nexo Whisper does not intermediate between you and the AI. You connect directly to the provider you already have. We don't charge for tokens or see your data.

Available providers

Anthropic (Claude)

  1. Go to console.anthropic.com/keys.
  2. Create a new API key.
  3. In Nexo Whisper → Settings → AI Enhancement → Provider → choose Anthropic.
  4. Paste the key in the corresponding field.
  5. Choose the model: claude-haiku-3-5 (fast and economical) or claude-sonnet-4-5 (better quality).

Which model to choose?

For daily dictation, claude-haiku-3-5 is the best speed/cost ratio. claude-sonnet-4-5 for more complex texts where quality matters more.

OpenAI (GPT)

  1. Go to platform.openai.com/api-keys.
  2. Create an API key.
  3. In Nexo Whisper → Settings → AI Enhancement → Provider → choose OpenAI.
  4. Paste the key.
  5. Recommended model: gpt-4o-mini (economical) or gpt-4o (better quality).

Google Gemini

  1. Go to aistudio.google.com → Get API Key.
  2. In Nexo Whisper → Settings → AI Enhancement → Provider → choose Google Gemini.
  3. Paste the key.
  4. Recommended model: gemini-2.0-flash.

Groq (ultra-fast cloud)

Groq is the fastest option for AI enhancement — responds in under a second.

  1. Go to console.groq.com/keys.
  2. Create an API key (free plan available).
  3. In Nexo Whisper → Provider → Groq.
  4. Paste the key.
  5. Recommended model: llama-3.1-8b-instant.

Local CLIs (automatic detection)

If you already have an AI CLI installed on your Mac, Nexo Whisper detects it automatically:

| CLI | Detection | |-----|-----------| | Claude Code (claude) | Automatic | | OpenAI Codex (codex) | Automatic | | Ollama (local models) | Automatic if running |

When a CLI is detected, it appears as an option in the provider selector. No API key needed.

100% local processing with Ollama

If you use Ollama with a local model (e.g.: llama3.2:3b), AI enhancement also runs on your Mac. Slower than cloud but completely private.

Custom OpenAI-compatible

If you use a proxy, OpenRouter, or any service compatible with the OpenAI API:

  1. Provider → Custom (OpenAI-compatible).
  2. Enter the Base URL of the endpoint.
  3. Enter the API key.
  4. Enter the model name.

Enable and disable enhancement

You have several options:

  • Menu bar: click the Nexo Whisper icon → toggle "AI Enhancement".
  • Keyboard shortcut: the shortcut configured in Settings → Enhancement shortcuts.
  • Settings: toggle in the main settings screen.

View token spend

Nexo Whisper tracks estimated spend in Settings → Dashboard. There you can see:

  • Tokens used this month by provider.
  • Estimated cost in USD.
  • Breakdown by model.

Next steps

Anterior
Streaming en vivo
Siguiente
Prompts predefinidos