A dynamic analytical tool that cross-references BenchLM coding leaderboards with OpenRouter pricing to identify the best AI models for software development.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install top-coding-models
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install top-coding-models using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Top Coding Models?
The Top Coding Models skill is a technical powerhouse designed for developers and architects who need real-time data to select the most effective LLMs for their coding workflows. By integrating with Openclaw Skills, this tool fetches live benchmark data from BenchLM—weighted 50/50 between SWE-bench Pro and LiveCodeBench—to ensure users are making decisions based on actual performance rather than marketing hype.
Beyond simple rankings, the skill provides a comprehensive cost-benefit analysis by pulling real-time token pricing and context window limits from the OpenRouter API. This ensures that whether you are looking for the absolute highest quality or the best budget-friendly open-weight model, you have the data needed to optimize your AI-assisted development environment.
Top Coding Models Use Cases
- Comparing SWE-bench Pro scores against real-time OpenRouter pricing to find the highest value models.
- Identifying free-tier models on OpenRouter for low-cost development and testing.
- Verifying model compatibility and tool-calling support for agentic IDEs like Claude Code, Cursor, and Windsurf.
- Checking maximum context windows and output token limits for large-scale codebase refactoring.
- Fetching exact OpenRouter model IDs for seamless API integration into custom coding agents.
How Top Coding Models Works
- The skill triggers a GET request to the BenchLM API to retrieve the latest coding leaderboard rankings and scores.
- It concurrently fetches the latest model metadata, including live pricing and context limits, from the OpenRouter Models API.
- A fuzzy matching algorithm normalizes model names from BenchLM to map them accurately to OpenRouter IDs.
- The system calculates a cost-performance ratio, identifying 'Best Value' and 'Free' models.
- A structured Markdown table or JSON object is generated, detailing rank, score, pricing, and IDE compatibility.
Top Coding Models Setup
To install and run this skill within your Openclaw Skills directory, follow these steps:
# Navigate to the scripts directory
cd ~/.config/opencode/skills/top-coding-models/scripts/
# Run the fetcher to see the top 20 models
python3 fetch_models.py
# To get machine-readable JSON output for integrations
python3 fetch_models.py --json
# To limit results to a specific number of top models
python3 fetch_models.py --top=10
Top Coding Models Data Schema & Taxonomy
The skill organizes its output into a highly structured format, ensuring compatibility with both human readers and automated systems.
| Attribute | Description |
|---|---|
| Rank | Current position on the BenchLM leaderboard. |
| Model ID | The specific OpenRouter identifier for API calls. |
| Coding Score | 50/50 weighted average of SWE-bench and LiveCodeBench. |
| Pricing | Input and Output costs normalized per 1 million tokens. |
| Capabilities | Indicators for Tool Calling and Structured Output support. |
| Context | Maximum context window and output token limits. |
Top Coding Models Advanced Features
- Real-time API Integration: Bypasses caching to ensure data is never stale, pulling directly from BenchLM and OpenRouter on every execution.
- Agentic IDE Matrix: Specifically tracks compatibility for advanced coding agents like Aider, Cline, and Roo Code.
- Automated Cost-Performance Analysis: Silently calculates the best value per benchmark point to assist in budget optimization.
- Flexible Output Formats: Supports both a visual Markdown table for users and a strict JSON schema for multi-agent orchestration within Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Chat Record Generator
A sophisticated tool for generating synthetic, multi-agent group chat records in Excel format for model training and simulation.
Datayes Macro
A professional-grade interface for querying Datayes and Tonglian macro-economic and industry indicators.
ClawTrace
A diagnostic utility that analyzes OpenClaw logs and session context to generate detailed performance reports and optimization insights.
Image Prompt Engineer
A technical prompt engineering framework designed to translate visual concepts into high-precision technical instructions for AI image generators.
Teacher Lesson Plans & Assessment Generator
A comprehensive pedagogical assistant for UK teachers to generate curriculum-aligned lesson plans, worksheets, and student reports.
Gothic Portrait Generator
A specialized AI image generation skill for creating high-fidelity gothic portraits with dramatic lighting and Victorian aesthetics.
Featured*
