How to Configure Your Own AI Provider in Falcon PDF

Connect OpenAI, OpenRouter, Groq, Together AI, or a local Ollama model and chat privately with your PDFs using your own API key.

Falcon PDF Team··6 min read
Table of contents

Falcon PDF AI Chat lets you ask questions about your document, summarize pages, highlight text, and more. By default it uses the built-in Falcon AI provider. If you want unlimited usage, more powerful models, or full privacy, you can plug in your own AI provider — either a cloud API (OpenAI, OpenRouter, Groq, Together AI) or a local model via Ollama.

Bring your own key

When you use a Custom or Local provider, your requests go directly from the app to that provider using your API key. Falcon PDF never sees your key beyond storing it locally on your device.

Step 1: Open AI Chat settings

  1. Open any PDF in Falcon PDF.
  2. Open AI Chat. On mobile, tap the assistant icon then the gear ⚙️ in the top bar. On desktop, open the AI Chat panel and click the gear ⚙️ in the chat header.
  3. You'll see three provider options: Falcon AI, Local Model (Ollama) (desktop), and Custom Provider.

Step 2: Choose a custom provider

Select Custom Provider. Pick a preset to auto-fill the base URL, then paste your API key and model name.

ProviderBase URLExample model
OpenAIhttps://api.openai.com/v1gpt-4o-mini
OpenRouterhttps://openrouter.ai/api/v1openai/gpt-4o-mini
Groqhttps://api.groq.com/openai/v1llama-3.3-70b-versatile
Together AIhttps://api.together.xyz/v1meta-llama/Llama-3.3-70B-Instruct-Turbo

Getting an API key

Tip

The model must support tool / function calling for actions like highlight and search to work. We recommend gpt-4o-mini or a 70B-class model.

  1. Select the preset (e.g. OpenAI). The base URL fills in automatically.
  2. Paste your API key.
  3. Enter the model name (see examples above).
  4. Tap Save. The new provider takes effect immediately.

Step 3 (Desktop): Use a local model with Ollama

Prefer 100% private, unlimited chat with no API bills? Run a model locally with Ollama. (Local models are available on desktop.)

  1. Install Ollama and pull a capable model, for example:
ollama pull qwen2.5:14b
  1. Make sure Ollama is running (it serves on http://localhost:11434 by default).
  2. In Falcon PDF, choose Local Model (Ollama), confirm the endpoint, then tap Refresh to load your installed models and pick one.
  3. Tap Save. Chat now runs entirely on your machine.

Troubleshooting

  • “Failed to get AI response”: double-check the base URL, API key, and exact model name. A wrong model name is the most common cause.
  • Actions (highlight/search) don't run:your model likely doesn't support tool calling. Switch to a stronger model.
  • Ollama not detected: confirm Ollama is running and the endpoint is http://localhost:11434, then tap Refresh.
  • Billing errors:add credits/billing on your provider's dashboard.

FAQ

Is my API key safe?Yes — it's stored locally on your device and used only to talk to the provider you chose.

Do custom/local providers cost extra in Falcon PDF? No. You pay only your chosen provider (or nothing at all for local Ollama).

Which is best? OpenRouter is the easiest for trying many models, Groq is the fastest, and Ollama is the most private. Once configured, try using AI to annotate your PDF or merge documents before you chat.