Skip to content

AI_Model_Options.md

Latest commit

 

History

History
228 lines (161 loc) · 5.98 KB

AI_Model_Options.md

File metadata and controls

228 lines (161 loc) · 5.98 KB

AI Model Options for Dex

Quick setup: Run /ai-setup in Dex — I'll guide you through everything.

Overview

Dex can use different AI models depending on your needs:

Option Cost Internet Required Best For
Claude (default) ~$3-15/M tokens Yes Best quality, complex tasks
Budget Cloud ~$0.14-3/M tokens Yes Daily tasks, save 80%+
Offline Free No Travel, privacy

Option 1: Budget Cloud Models

What Are They?

Other companies make AI models that cost much less than Claude:

  • Kimi K2.5 (Moonshot AI) — 80% cheaper, similar quality
  • DeepSeek V3 — 95% cheaper, great for coding
  • Gemini Flash (Google) — 97% cheaper, handles long documents

Why Use Them?

  • Save money on everyday tasks
  • Claude is overkill for simple stuff
  • Still get excellent results for most things

Quality Comparison

Task Claude Kimi DeepSeek
Complex planning ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ⭐⭐⭐
Meeting summaries ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ⭐⭐⭐⭐
Daily planning ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ⭐⭐⭐⭐
Code help ⭐⭐⭐⭐⭐ ⭐⭐⭐ ⭐⭐⭐⭐⭐
Creative writing ⭐⭐⭐⭐⭐ ⭐⭐⭐ ⭐⭐⭐

How to Set Up

  1. Run /ai-setup in Dex
  2. Choose "Budget Cloud Mode"
  3. Follow the prompts to create an OpenRouter account
  4. Add $5-10 in credits (lasts weeks or months)
  5. Done! Use /model to switch between models

Cost Calculator

Usage Level Claude/Month With Budget Models
Light (10 queries/day) ~$15 ~$3
Medium (30 queries/day) ~$45 ~$9
Heavy (100 queries/day) ~$150 ~$30

Option 2: Offline Mode

What Is It?

Download an AI model directly to your computer. It runs without internet.

Why Use It?

  • ✈️ Works on airplanes
  • 🚂 Works on trains with bad WiFi
  • 🔒 Your data never leaves your computer
  • 💰 Completely free forever

Requirements

Your RAM Recommended Model Quality
8 GB Qwen 2.5 7B Good for basic tasks
16 GB Qwen 2.5 14B Good for most tasks
32 GB+ Qwen 2.5 32B Great for everything

Not sure about your RAM?

  • Mac: Apple menu → About This Mac → look for "Memory"
  • Windows: Right-click Start → System → look for "Installed RAM"

How to Set Up

  1. Run /ai-setup in Dex
  2. Choose "Offline Mode"
  3. Answer questions about your computer
  4. Install Ollama (free app) from ollama.ai
  5. Download the recommended model
  6. Done! Works automatically when offline

What Works Offline

Works great:

  • Daily planning and reviews
  • Meeting summaries
  • Task management
  • Note organization
  • Quick questions

Better with Claude:

  • Quarterly planning
  • Career coaching
  • Complex analysis
  • Creative writing

Option 3: Smart Routing

What Is It?

Let Dex automatically pick the best model for each task:

  • Complex stuff → Claude (best quality)
  • Simple stuff → Budget model (saves money)
  • No internet → Local model (offline backup)

How It Works

Task Model Used Reason
/quarter-plan Claude Needs deep thinking
/daily-plan Budget (Kimi) Routine task
/meeting-prep Budget (Kimi) Straightforward
/career-coach Claude Needs nuance
"What's on my calendar?" Budget Simple question
No internet Local Only option

How to Enable

  1. Run /ai-setup in Dex
  2. Set up budget cloud first
  3. Optionally set up offline mode
  4. Choose "Smart Routing"
  5. Done! I'll pick automatically (you can always override)

Quick Reference

Switching Models

Method 1: Ask me

  • "Use the budget model for this"
  • "Switch to Kimi"
  • "Use Claude for this one"

Method 2: Model picker

  • Type /model in pi
  • Select from the list

Checking Your Setup

Run /ai-status to see:

  • Which models are configured
  • Current model in use
  • OpenRouter credit balance
  • Ollama status

Model Names

Display Name Technical ID
Claude Sonnet claude-sonnet-4-20250514
Kimi K2.5 moonshotai/kimi-k2.5
DeepSeek V3 deepseek/deepseek-chat
Gemini Flash google/gemini-2.0-flash-exp:free
Qwen (Offline) qwen2.5:14b

Troubleshooting

Budget model not working?

  1. Check OpenRouter has credits (openrouter.ai/credits)
  2. Verify API key is correct (starts with sk-or-)
  3. Try regenerating the key

Offline model too slow?

Your model might be too big for your RAM:

  1. Open Activity Monitor (Mac) or Task Manager (Windows)
  2. If RAM is maxed, download a smaller model: 
    ollama pull qwen2.5:7b

Ollama won't start?

Mac:

  • Open Applications, right-click Ollama → Open
  • If blocked: System Settings → Privacy & Security → Allow

Manual start:

  • Open Terminal
  • Run: ollama serve
  • Keep window open

FAQ

Q: Will I notice a quality difference with budget models? A: For everyday tasks (planning, summaries, organization), most people don't notice. For complex reasoning or creative work, Claude is noticeably better.

Q: How much will I actually save? A: Depends on usage, but typically 50-80% if you use budget models for routine tasks and Claude for complex ones.

Q: Is my data safe with other models? A: Budget cloud models (Kimi, DeepSeek) are hosted by other companies. If privacy is critical, use offline mode — data never leaves your computer.

Q: Can I go back to Claude-only? A: Yes! Just don't use the other models, or disable them in settings.

Q: What if I'm offline and haven't set up local models? A: Dex will let you know and offer to set it up when you're back online. You can still browse/edit files manually.

Related