Overview
The Research Memex methodology supports both GUI and CLI (command-line interface) approaches to agentic AI workflows. This guide helps you choose the right tools for your research needs.
GUI vs CLI: GUI tools like Cherry Studio are visual and beginner-friendly. CLI tools offer power, automation, and scriptability. Most researchers use BOTH - GUI for exploration, CLI for production workflows.
GUI: Cherry Studio
- Visual interface with buttons, panels, menus
- Multi-model access through one app
- Knowledge bases for PDF analysis
- Beginner-friendly with no terminal needed
- Full guide: Cherry Studio Setup
CLI: Three Powerful Options
1. Claude Code (Premium)
- Provider: Anthropic (paid)
- Strength: Best reasoning quality
- Cost: ~$20/mo
- Full guide: Claude Code Setup
2. Gemini CLI (Free Power)
- Provider: Google (free tier!)
- Strength: 1M context, 1K req/day free
- Cost: FREE (or pay for more)
- Full guide: Gemini CLI Setup
3. OpenCode (Multi-Provider)
- Provider: 75+ options (community)
- Strength: Model flexibility, beautiful TUI
- Cost: Pay per provider
- Full guide: OpenCode Setup
Comprehensive Comparison Table
Feature Comparison
Feature | Cherry Studio | Claude Code | Gemini CLI | OpenCode |
---|
Interface | GUI | CLI | CLI | TUI (visual CLI) |
Model Access | 100+ | Claude only | Gemini only | 75+ |
Context Window | Varies | 200K | 1M | Varies |
Free Tier | Via providers | ❌ No | ✅ Yes (1K/day) | Via providers |
Cost | Free app | $20/mo | FREE | Pay per provider |
MCP Support | ✅ Full | ✅ Full | ✅ Full | ✅ Full |
Project Memory | Knowledge Base | CLAUDE.md | GEMINI.md | Config files |
Slash Commands | ❌ No | ✅ Yes | ✅ Yes | ⚠️ Limited |
Multi-Session | ✅ Tabs | ❌ Single | ❌ Single | ✅ Yes |
Git Integration | ❌ No | ✅ Built-in | ⚠️ Basic | ⚠️ Basic |
Open Source | ✅ Apache 2.0 | ❌ No | ✅ Apache 2.0 | ✅ MIT |
Best For | Exploration | Premium work | Budget + power | Flexibility |
Cost Comparison
Monthly cost estimates (heavy research use):
Tool | Setup Cost | Monthly Cost | Notes |
---|
Cherry Studio | Free | $5-30 | Via API providers |
Claude Code | Free | $20+ | Subscription + API |
Gemini CLI | Free | $0-10 | Free tier often enough! |
OpenCode | Free | $5-50 | Depends on providers chosen |
Budget strategy:
- $0/month: Gemini CLI + Cherry Studio (free tiers)
- $20/month: Add Claude Code for premium reasoning
- $30/month: All tools + OpenCode with multiple providers
By Research Phase
Phase 1: Exploration & Learning
→ Cherry Studio (visual, easy, multi-model testing)
Phase 2: Literature Analysis
→ Gemini CLI (1M context for 50+ papers, FREE!)
Phase 3: Automation & Scale
→ OpenCode (multi-provider, cost optimization)
Phase 4: Final Theory/Writing
→ Claude Code (premium reasoning quality)
By Use Case
“I need to analyze 50+ papers comprehensively”
→ Gemini CLI (1M context window)
“I want to compare GPT vs Claude vs Gemini vs DeepSeek”
→ OpenCode (75+ providers)
“I need the absolute best reasoning quality”
→ Claude Code (Anthropic Sonnet 4.5)
“I’m on a tight budget”
→ Gemini CLI (1K free requests/day) + Cherry Studio (free app)
“I prefer visual interfaces”
→ Cherry Studio (GUI)
“I want to automate everything”
→ Any CLI (all support automation)
“I need maximum privacy (local processing)”
→ OpenCode + Ollama (run models locally)
By Skill Level
Beginner:
- Start: Cherry Studio (GUI, no terminal needed)
- Then: Gemini CLI (gentle CLI introduction, free!)
Intermediate:
- Add: OpenCode (experiment with providers)
- Or: Claude Code (premium quality)
Advanced:
- Use all 4 tools strategically
- Right tool for each task
- Optimize cost-quality-speed tradeoff
Strategy 1: Budget Maximizer (Free/Minimal Cost)
Tools:
- Cherry Studio (free app)
- Gemini CLI (free 1K req/day)
Workflow:
- Cherry Studio: Exploration, testing, learning
- Gemini CLI: Heavy analysis (50+ papers), automation
Cost: $0-5/month
Strategy 2: Quality Optimizer (Premium)
Tools:
- Cherry Studio (multi-model testing)
- Claude Code (premium reasoning)
- Gemini CLI (massive context backup)
Workflow:
- Cherry Studio: Quick tests, model comparison
- Gemini CLI: Large-scale synthesis (free!)
- Claude Code: Final theoretical work (paid quality)
Cost: $20-30/month
Strategy 3: Model Experimenter (Flexibility)
Tools:
- OpenCode (75+ providers)
- Cherry Studio (visual multi-model)
- Gemini CLI (free Google power)
Workflow:
- OpenCode: Test 5+ models on same task
- Cherry Studio: Visual comparison
- Gemini CLI: Final choice if Google wins
Cost: $10-40/month (depends on experimentation)
Strategy 4: Research Lab (Complete)
Tools:
Workflow:
- Cherry Studio: Student onboarding, teaching
- Gemini CLI: RA heavy lifting (free!)
- OpenCode: Model experiments, optimization
- Claude Code: PI final review, theory building
Cost: $20-50/month
Getting Started Recommendation
For Course Students (Recommended Path)
Week 1-2: Install Cherry Studio
- Learn GUI basics
- Test multiple models
- Build knowledge bases
Week 3-4: Add Gemini CLI
- Learn terminal basics
- Try massive context (30+ papers)
- Automate screening/extraction
Week 5+: Add OpenCode OR Claude Code
- OpenCode if: Want flexibility, tight budget
- Claude Code if: Need premium quality, budget allows
Not required: All 3 CLIs
Recommended: Cherry Studio + At least 1 CLI (Gemini preferred for free tier)
CLI Basics for Researchers
Terminal Fundamentals
If you’re new to terminals:
Navigate to project:
cd ~/Documents/Research-Project # Change directory
ls # List files
pwd # Show current location
mkdir new-folder # Create folder
Launch CLI tools:
claude # Launch Claude Code
gemini # Launch Gemini CLI
opencode # Launch OpenCode
Exit:
Environment Variables
What are they?
Variables that store API keys securely.
Set temporarily (current session):
export GOOGLE_API_KEY="AIza..."
Set permanently:
Add to ~/.zshrc
(macOS) or ~/.bashrc
(Linux):
echo 'export GOOGLE_API_KEY="AIza..."' >> ~/.zshrc
source ~/.zshrc
Check if set:
Troubleshooting
”I’ve never used terminal before”
Resources:
Or stick with Cherry Studio (GUI) - it’s powerful enough!
Decision tree:
-
Comfortable with terminal?
- No → Cherry Studio
- Yes → Continue
-
Budget?
- $0 → Gemini CLI
- $20+ → Claude Code
- Flexible → OpenCode
-
Need massive context (50+ papers)?
- Yes → Gemini CLI (1M context)
- No → Any tool works
Absolutely! Most researchers use:
- GUI for exploration
- CLI for automation
- Both for different phases
Next Steps
1. Choose your tool(s) using the decision matrix above
2. Follow detailed setup guide:
3. Learn agentic workflows:
4. Connect to research tools:
Resources
Comparison Articles
Official Documentation
Video Tutorials
FAQ
Q: Do I need a CLI tool or is Cherry Studio enough?
A: Cherry Studio is sufficient for the course! CLI tools are optional for students who want automation, scripting, or prefer terminal workflows.
Q: Which CLI tool is best?
A: Depends on priority:
- Budget → Gemini CLI (free)
- Quality → Claude Code (premium)
- Flexibility → OpenCode (75+ providers)
Q: Can I switch between tools?
A: Yes! All use markdown files and similar workflows. Easy to migrate.
Q: Do these work with the same Zotero setup?
A: Yes! All connect to your Zotero library via MCP servers configured in the Zotero Setup Guide.
Q: Are CLI tools harder to learn?
A: Initial learning curve, but automation benefits compound over time. Gemini CLI or OpenCode’s TUI are good starting points.