LLM Selection Skill

A comprehensive decision-making framework for selecting the most suitable Large Language Model (LLM) based on your specific use cases, budget constraints, and compliance requirements.

🌟 Features

Capability	Description
Scenario Analysis	10+ scenario types (chat, coding, creative writing, reasoning, multimodal, etc.)
Global & Chinese Models	OpenAI/Claude/Gemini + Hunyuan/Qwen/ERNIE/DeepSeek/GLM/Kimi/Doubao
Hard Constraint Filtering	Data compliance, context length, deployment mode, multimodal support
Scoring Matrix	Reasoning 30% + Multimodal 25% + Context 20% + Speed 15% + Cost 10%
Cost Estimation	Automatically calculate monthly costs based on usage volume
Implementation Guide	Canary testing plan + Fallback strategy

📁 File Structure

llm-selection/
├── SKILL.md                    ← Skill entry point
├── references/
│   ├── model_zoo.md           ← 30+ mainstream model comparisons
│   └── selection_framework.md ← Selection framework, scoring matrix, scenario cheat sheet
└── scripts/
    └── cost_estimator.py      ← Cost estimation script

🚀 Installation

Run the following command to install the skill:

npx skills add https://github.com/dimayip/llm-selection.git

💡 Usage

Trigger the skill by using keywords in your CodeBuddy conversation:

• "Help me choose a model"
• "Which model for [scenario]"
• "Compare Model A and Model B"
• "LLM selection"
• "Cost estimation"

The skill will automatically启动 a 5-step selection process:

Step 1: Requirement Collection

• Ask about: use case, budget, data compliance requirements
• Optional: context length, QPS, deployment preference, multimodal needs

Step 2: Hard Constraint Filtering

• Data compliance → Only Chinese models for domestic business
• Context length → Filter out models that don't meet requirements
• Deployment mode → API / Can be local / Must be local

Step 3: Scoring & Ranking

Weighted scoring:

• Task match 40% + Cost 30% + Latency 15% + Stability 10% + Usability 5%

Step 4: Selection Report

Output includes:

• Top 3 recommendations (Primary / Alternative / Best Value)
• Cost comparison table
• Implementation suggestions (canary testing + fallback strategy)

Step 5: Cost Estimation

Run the cost estimator:

python3 scripts/cost_estimator.py

🌐 Supported Models

International Models

• OpenAI: GPT-4o, GPT-4o-mini, o1, o3
• Anthropic: Claude 3.5 Sonnet, Claude 3.7 Sonnet, Claude 4 Sonnet
• Google: Gemini 2.0 Flash, Gemini 2.5 Pro

Chinese Models

• Tencent Hunyuan: Hunyuan-Turbo, Hunyuan-Pro, Hunyuan-T1
• Alibaba Qwen: Qwen-Max, Qwen-Plus, Qwen-Turbo, Qwen3-235B
• Baidu ERNIE: ERNIE-4.5-Turbo, ERNIE-Spark
• DeepSeek: DeepSeek-V3, DeepSeek-R1
• Zhipu GLM: GLM-4-Plus, GLM-4-Air
• Moonshot Kimi: Moonshot-v1-128k
• ByteDance Doubao: Doubao-Pro-32k

📊 Cost Estimation

The scripts/cost_estimator.py script helps you estimate monthly costs:

python3 scripts/cost_estimator.py [daily_calls] [avg_input_tokens] [avg_output_tokens]

Example:

python3 scripts/cost_estimator.py 10000 500 300

This will output a comparison table with estimated monthly costs for all supported models.

🔍 Selection Framework

For detailed information about the selection framework, scoring matrix, and scenario-specific recommendations, see:

• Model Zoo - Detailed comparison data for 30+ mainstream models
• Selection Framework - Scoring matrix, scenario cheat sheet, and implementation guide

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📧 Contact

• GitHub Issues: https://github.com/dimayip/llm-selection/issues
• Author: bellchen

---

⭐ If you find this skill useful, please consider giving it a star on GitHub!