update-models

name	update-models
description	Update model data in lib/models.ts with latest specifications from OpenAI and Google APIs

Update Model Data Skill

This skill updates the model specifications in lib/models.ts with the latest data from OpenAI and Google (Gemini) APIs.

Instructions

When invoked, follow these steps:

1. Read Current Model Data

First, read the current lib/models.ts file to understand the existing structure and models.

2. Fetch Latest OpenAI Model Information

Search for and fetch the latest OpenAI model specifications:

• Model names and IDs
• Context windows (max input and output tokens)
• Pricing per million tokens (input, cached input, output)
• Model descriptions and capabilities
• Knowledge cutoff dates
• Any deprecation notices

Focus on these model families:

• GPT-4o and GPT-4o-mini
• o-series (o1, o3, o4-mini, etc.)
• GPT-4.1
• Any new production-ready models

3. Fetch Latest Google Gemini Model Information

Search for and fetch the latest Google Gemini model specifications:

• Model names and IDs
• Context windows (max input and output tokens)
• Pricing per million tokens/characters (input and output)
• Model descriptions and features
• Tiered pricing information (if applicable)
• Special features (context caching, batch API, etc.)
• Any deprecation notices

Focus on these model families:

• Gemini 3 Pro and Flash
• Gemini 2.5 Pro, Flash, and Flash-Lite
• Gemini 2.0 models
• Image generation models (Gemini Pro Image, etc.)
• Any new production-ready models

4. Update the Models File

For each model that needs updating:

• Update pricing if it has changed
• Update context windows if they've changed
• Update descriptions with new capabilities
• Add new models that are production-ready
• Mark deprecated models with disabled: true and add deprecation notices
• Ensure all pricing is in dollars per million tokens
• Verify maxInputTokens and maxOutputTokens are accurate

5. Fetch performance data

If available, fetch any performance benchmarks or latency information for the models to set the outputTokensPerSecond field on models. Here's a site that provides this data: https://artificialanalysis.ai/leaderboards/models

6. Verify Changes

After making updates:

• Read the updated sections to verify correctness
• Ensure TypeScript syntax is valid
• Check that new models follow the existing type structure
• Verify that pricing information is accurate

7. Provide Summary

Create a clear summary for the user showing:

• Which models were updated
• What changes were made (pricing, context windows, descriptions)
• Which models were added
• Which models were deprecated
• Links to official documentation sources

Important Notes

• Always cite official sources (OpenAI and Google documentation)
• Use WebSearch to find current documentation pages
• Prefer official API documentation over third-party sources
• When WebFetch fails, use multiple search queries to gather complete information
• Be conservative with updates - only change what's clearly documented
• Include sources at the end of your response as markdown links
• For models with tiered pricing (e.g., different rates above 200k tokens), note this in the description
• Update both text and image models if applicable

Example Usage

User can invoke this skill by typing:

/update-models

or

Please update the model data