AI Providers

Set up Claude, GPT, Gemini, Groq, and more for AI text processing.

VivaDicta supports 20+ AI providers for text processing. You can use API keys, sign in with an existing subscription, run models locally on your network, or download a model and run AI fully on your device.

Cloud Providers

ProviderFree TierTop ModelsAPI Key
GroqFree foreverLlama modelsGet key
Google GeminiFree tier availableGemini 3 Pro/FlashGet key
MistralFree tier, rate-limited. No cardmistral-medium, VoxtralGet key
Cerebras$5 credits. Card needed to issue a key, removable afterqwen-3.8-27b, gpt-oss-120bGet key
xAI (Grok)Free credits for new accountsGrok modelsGet key
GitHub CopilotFree tier availableClaude, GPT, GeminiSign in via app
OpenRouterSome free modelsMany models availableGet key
OpenCode Zen & GoFree models on ZenClaude, GPT-5, GLM, Kimi, QwenGet key
Ollama CloudFree tier, then a subscriptiongpt-oss, DeepSeek, Qwen, GLMGet key
OllamaCompletely free (local)Llama, Mistral, GemmaNo key needed
Apple Foundation ModelFree (on-device)Apple on-device modelNo key needed
On-Device ModelsFree (on-device)Gemma, Qwen, Phi (download)No key needed

Setup Pattern

Most cloud providers follow the same setup flow:

  1. Create an account with the provider and get an API key.
  2. Open VivaDicta → Settings → select your mode → AI Processing.
  3. Choose the provider and paste your API key.
  4. Select a model from the available list.

Ollama (Local, Free)

Ollama runs AI models locally on a Mac or PC on your network. It's completely free and private — no data leaves your local network.

  1. Install Ollama on a Mac or PC on your local network.
  2. Pull a model (e.g., ollama pull llama3.2).
  3. In VivaDicta, select Ollama as your AI provider and enter your computer's local IP address.

Custom OpenAI-Compatible

Any endpoint that speaks the OpenAI chat completions API format works with VivaDicta. Enter the endpoint URL and optional API key in the AI processing settings.

Use Your Existing Subscriptions

If you have ChatGPT Plus, GitHub Copilot (free tier available), or Claude Pro, you can use VivaDicta's AI features at no extra cost. See Use Your Subscriptions.

Which Provider Should I Choose?

  • Best free option: Cerebras — qwen-3.8-27b scores 9.2 and answers in 0.3s, quicker than anything else measured. Issuing a key asks for a card and gives you $5 of credits; you can remove the card straight after. Ollama Cloud's gemma4:31b and Groq are the other free picks.
  • Best free option without a card: Mistral — mistral-medium-latest scores higher at 9.4 and needs nothing but an account, but its free tier is rate-limited hard enough that you wait several seconds between dictations.
  • Best quality: Google Gemini — gemini-3.8-flash tops the model ranking at 9.7, still under a second.
  • Best value with subscription: GitHub Copilot — free tier available, or $10/mo for Pro with access to Claude, GPT, Gemini, and Grok.
  • Maximum privacy: a downloadable on-device model (Gemma, Qwen, Phi - no iOS 26 needed), Apple Foundation Model (requires iOS 26+), or Ollama - everything stays on your device or local network.

See Free AI Tiers for a detailed breakdown of free options.