Skip to main content

​
Contributing to Open Research

We welcome contributions from the community! This guide explains how to add model discoveries, research papers, and applied tasks to our database.

​
Ways to Contribute

Model Discoveries

Found a new AI model on social media? Add it!

Research Papers

Share papers that impact model performance

Prompt Examples

Viral prompts with engagement data

Applied Tasks

Document problem-solving workflows

​
Model Discoveries

​
What We Track

When you discover a new AI model, document: 
Required:
  - name: "Model Name"
  - provider: "OpenAI | Anthropic | Google | etc"
  - source_url: "Link to announcement"
  - platform: "x | instagram | linkedin | arxiv"
  - posted_date: "YYYY-MM-DD"
  
Optional but valuable:
  - engagement:
      likes: 45000
      shares: 12000
      views: 2500000
      bookmarks: 8900
  - capabilities: ["chat", "vision", "audio"]
  - description: "What the model does"

​
How to Submit

  1. Fork the repository
  2. Create a new file in research/models/ using the template
  3. Fill in the frontmatter with all available data
  4. Add social proof - screenshots or links to announcements
  5. Submit a PR with description of discovery source

​
Template

---
title: "Model Name"
description: "Brief description"
---

# Model Name

<Note>
  **Status**: 🔵 Discovered  
  **Source**: X/Twitter | LinkedIn | Instagram | arXiv
</Note>

## Overview

| Property | Value |
|----------|-------|
| **Provider** | Company Name |
| **Release Date** | YYYY-MM-DD |
| **Modalities** | Text, Image, Audio |

## Discovery Source

<Card>
  **Platform**: X (Twitter)  
  **Author**: @username  
  **Posted**: Date  
  **Engagement**: X likes | Y shares | Z views | W bookmarks
  
  > "Quote from the announcement"
  
  [View Original →](link)
</Card>

## Claims to Validate

- [ ] Context window size
- [ ] Multimodal capabilities
- [ ] Benchmark scores
- [ ] Pricing information

## Related Models

- [Related Model](/research/models/related-model)

​
Research Papers

​
What Makes a Good Paper Contribution

We prioritize papers that:
  • Directly impact deployed models (e.g., improves GPT-4o by X%)
  • Introduce new benchmarks or evaluation methods
  • Challenge existing assumptions (negative results welcome!)
  • Have reproducible results with available code

​
Impact Levels

Paradigm shift. Examples: Transformers, GPT-3 moment, breakthrough architecture
Major measurable improvement (>10% on key benchmark)
Small but meaningful improvement (2-10%)
Reproduces or confirms existing work
Challenges or disproves existing claims (valuable!)

​
Required Information

Required:
  - title: "Paper Title"
  - authors: ["Author Name"]
  - source: "arxiv | neurips | icml | etc"
  - source_id: "2501.12345"
  - url: "https://arxiv.org/abs/2501.12345"
  - impact_level: "significant"
  
For model impacts:
  - affected_models: ["gpt-4o", "claude-3-opus"]
  - benchmark_changes:
      math: { old: 76.4, new: 87.9 }

​
Applied Tasks

​
Task Contribution Format

Document real-world problem-solving workflows: 
Required:
  - title: "Task Name"
  - category: "content-creation | coding | analysis"
  - steps: ["Step 1", "Step 2", ...]
  - models_used: ["gpt-4o", "claude-3-opus"]
  
Optional:
  - cost: 0.47
  - quality_score: 8.5
  - studio: "Content Creation"

​
Task Structure

Each task should include:
  1. Problem Definition - What are we solving?
  2. Capability Mapping - What AI capabilities are needed?
  3. Model Selection - Which frontier models? Why?
  4. Step-by-Step - Actual prompts and outputs
  5. Results - Quality metrics, cost breakdown
  6. Insights - What did we learn?

​
Data Schema

All contributions follow our open schemas:

​
Review Process

1

Submit PR

Create a pull request with your contribution
2

Automated Checks

Schema validation and link checking
3

Community Review

Other contributors can comment and suggest
4

Graduate Review

For model validations and paper analysis
5

Merge

Approved contributions are merged and deployed

​
Discord Integration

Contributions feed into our Discord community:
  • Model discoveries → #new-models channel
  • Paper impacts → #research-updates channel
  • Task workflows → #studio-showcase channel

​
Attribution

All contributors are credited:
  • GitHub contributors list
  • Page footers (“Contributed by @username”)
  • Monthly contributor highlights

​
Code of Conduct

  • Be accurate - Verify claims before submitting
  • Be transparent - Disclose conflicts of interest
  • Be respectful - Constructive feedback only
  • Be open - Embrace negative results and corrections

​
Questions?

Thank you for helping democratize AI research!