Setup Guide

YouGist works with whichever AI provider you supply credentials for. This page walks you through getting an API key from each supported provider, or installing a free local model runner.

The three-step flow for any provider

  1. Get credentials. Either an API key from a cloud provider, or install Ollama / LM Studio and pull a model.
  2. Open YouGist Options. Click the YouGist toolbar icon, then the ⚙ gear button in the side panel.
  3. Pick your provider from the dropdown, paste your key (or set the local URL + model), and click Save. That's it — go summarize a video.

Cloud providers

All cloud providers below use the same flow: visit their console, generate an API key, paste it into YouGist's Options page. Your key stays in your browser — it never touches YouGist's servers, and it goes directly from your browser to that provider's official API.

OpenAI

Default model: gpt-4o-mini · Pricing: usage-based · Get API key →

The most widely-used commercial AI API. Reliable, fast, well-documented.

  1. Open platform.openai.com/api-keys and sign in (or create an account; phone verification required).
  2. Click Create new secret key, give it a name (e.g. YouGist), and copy the value — it starts with sk-. You can't view it again after closing the dialog.
  3. In YouGist Options: set Active provider to OpenAI, paste the key into API Key, click Save.
Heads-up: OpenAI requires a payment method on file ($5 minimum prepaid credit). The default gpt-4o-mini costs roughly $0.0001 per video summary.

Anthropic

Default model: claude-haiku-4-5-20251001 · Pricing: usage-based · Get API key →

Claude models. Strong at long-context reasoning and following formatting instructions precisely.

  1. Open console.anthropic.com/settings/keys and sign in.
  2. Click Create Key, name it, copy the value (starts with sk-ant-).
  3. In YouGist Options: Active provider → Anthropic, paste the key. The default model field already contains the cheapest Haiku model — leave it or paste a different one from Anthropic's model list.
Heads-up: Requires prepaid credits ($5 minimum). Haiku is the cheapest and fastest tier — typically a few hundredths of a cent per summary.

Google Gemini

Default model: gemini-2.5-flash · Pricing: free tier available · Get API key →

Generous free tier. Good first choice if you don't want to set up billing anywhere.

  1. Open aistudio.google.com/apikey and sign in with a Google account.
  2. Click Create API key, pick a Google Cloud project (or let it create one), and copy the key — starts with AIza.
  3. In YouGist Options: Active provider → Google Gemini, paste the key. Use the recommended chip gemini-2.5-flash for the best free-tier balance.
Free tier: Rate-limited but generous enough for personal use. Hit the limit? Switch to gemini-2.5-flash-lite via the chip in Options.

xAI Grok

Default model: grok-4-1-fast-non-reasoning · Pricing: usage-based (free credits on signup) · Get API key →

Grok by xAI. Sign-up often comes with a free monthly credit allowance.

  1. Open console.x.ai and sign in with your X (Twitter) account.
  2. Go to API Keys → Create API Key, copy the value (starts with xai-).
  3. In YouGist Options: Active provider → xAI Grok, paste the key. The default grok-4-1-fast-non-reasoning is the fastest model — for higher quality, pick the reasoning chip.

DeepSeek

Default model: deepseek-v4-flash · Pricing: extremely low cost per token · Get API key →

Independent provider with very competitive pricing.

  1. Open platform.deepseek.com/api_keys and sign in.
  2. Click Create API Key, copy (starts with sk-).
  3. In YouGist Options: Active provider → DeepSeek, paste the key.
Note: The legacy deepseek-chat and deepseek-reasoner model names are deprecated on 2026-07-24. YouGist defaults to the current deepseek-v4-flash to avoid the breaking change.

Mistral AI

Default model: mistral-medium-3.5 · Pricing: usage-based · Get API key →

European AI provider, GDPR-friendly, strong multilingual coverage.

  1. Open console.mistral.ai/api-keys and sign in.
  2. Click Create new key, copy the value.
  3. In YouGist Options: Active provider → Mistral AI, paste the key.

Groq

Default model: llama-3.3-70b-versatile · Pricing: free tier, then usage-based · Get API key →

Famously fast inference for open-source models. Summaries often complete in 1–2 seconds.

  1. Open console.groq.com/keys and sign in (Google / GitHub auth).
  2. Click Create API Key, copy (starts with gsk_).
  3. In YouGist Options: Active provider → Groq, paste the key.
Free tier: Reasonable daily limits — perfect for personal use without billing setup.

OpenRouter

Default model: pick any from their catalog · Pricing: pay-as-you-go credits · Get API key →

Aggregator — one key, 100+ models from every major provider plus open-source. Great if you want to A/B test models or use exotic ones that don't have direct integration.

  1. Open openrouter.ai/keys and sign in (Google / GitHub).
  2. Top up credits, then click Create Key, copy (starts with sk-or-).
  3. Browse openrouter.ai/models and copy the slug of the model you want — e.g. anthropic/claude-sonnet-4.5 or meta-llama/llama-3.3-70b-instruct.
  4. In YouGist Options: Active provider → OpenRouter, paste the key, paste the model slug.

Local models

Run an open-source model on your own machine. Free, private, no API key. Requires enough RAM for the model (4–8 GB for a 7B-parameter model, more for bigger ones).

Ollama

Default URL: http://localhost:11434 · Recommended model: qwen2.5:7b · Install Ollama →

Command-line model runner. Tiny, fast to set up, runs on macOS, Linux, and Windows.

1. Install and start Ollama:

  1. Download from ollama.com/download and run the installer.
  2. Pull a model from your terminal: 
    ollama pull qwen2.5:7b
    This downloads ~5 GB. qwen2.5:7b is the recommended default — good multilingual coverage and fits on most laptops.
  3. Start Ollama with CORS enabled so the extension can talk to it: 
    OLLAMA_ORIGINS=* ollama serve
    (Without OLLAMA_ORIGINS=*, Ollama rejects requests from extensions.)

2. Configure in YouGist:

  1. Open YouGist Options.
  2. Active provider → Ollama.
  3. Server URL: http://localhost:11434 (already the default).
  4. Model: qwen2.5:7b — or click any chip to swap. Click Save.
Other model sizes: qwen2.5:3b for slower / low-RAM machines (faster, less detail); qwen2.5:14b for better quality if you have 16+ GB RAM. Pull the size you want with ollama pull qwen2.5:14b first.

LM Studio

Default URL: http://localhost:1234 · Recommended model: qwen/qwen3-4b-2507 · Install LM Studio →

GUI alternative to Ollama with a built-in model browser. Easier if you prefer clicking over typing.

1. Install and load a model:

  1. Download from lmstudio.ai and run the installer.
  2. Open LM Studio, click the search icon, find qwen/qwen3-4b-2507 (recommended: small, fast) and click Download.
  3. Once downloaded, click Load Model at the top.

2. Start the local server:

  1. Switch to the Developer tab (left sidebar).
  2. Click Start Server — defaults to port 1234.
  3. In server settings, make sure CORS is enabled so the extension can connect.

3. Configure in YouGist:

  1. Open YouGist Options.
  2. Active provider → LM Studio.
  3. Server URL: http://localhost:1234 (default).
  4. Model: paste the exact model identifier shown in LM Studio (e.g. qwen/qwen3-4b-2507) or click a recommended chip.
  5. API Key: leave as lm-studio (LM Studio ignores it but the OpenAI-compatible API requires a value).
  6. Save.
For better quality: Load qwen/qwen3.6-35b-a3b if you have 32+ GB RAM. It's a reasoning model — YouGist already sets max_tokens high enough to accommodate the thinking phase.