OpenAI Setup

​

Available Models

​

O-Series (Reasoning Models)

• o1 - Advanced reasoning with 200K context
• o3 - Enhanced reasoning capabilities
• o3-mini - Compact reasoning model
• o4-mini - Next-generation compact reasoning

​

GPT-5 Series (Latest)

• gpt-5 - Flagship model with 200K context
• gpt-5-mini - Compact version
• gpt-5-nano - Ultra-compact version
• gpt-5.1 - Enhanced GPT-5
• gpt-5.2 - Latest iteration
• gpt-5.3-codex - Optimized for coding (400K context)

​

GPT-4 Series

• gpt-4.1 - Enhanced GPT-4
• gpt-4o - Multimodal flagship (128K context)
• gpt-4o-mini - Fast and affordable
• gpt-4-turbo - Enhanced GPT-4
• chatgpt-4o-latest - Latest ChatGPT model

​

GPT-3.5 Series

• gpt-3.5-turbo - Fast and affordable (16K context)

• Tool calling and parallel execution
• Vision capabilities (GPT-4 and newer)
• Structured outputs
• Function calling

​

Setup Steps

forge provider login

model: gpt-4o

model: gpt-5.3-codex

forge

> Write a function to calculate fibonacci numbers

​

Configuration

​

API Endpoint

• Chat Completions: https://api.openai.com/v1/chat/completions
• Models List: https://api.openai.com/v1/models

​

Authentication

​

Model Selection Guide

​

For Coding and Development

• gpt-5.3-codex - 400K context, optimized for agentic coding
• gpt-4o - Excellent balance of capability and speed

• gpt-4o-mini - Fast and affordable
• gpt-3.5-turbo - Good for simple tasks

​

For Reasoning Tasks

• o3 - Advanced reasoning
• o1 - Strong reasoning capabilities

• o3-mini - Compact but capable
• o4-mini - Next-generation efficiency

​

For General Use

• gpt-5 - 200K context, most advanced
• chatgpt-4o-latest - Continuously updated

• gpt-4.1 - Enhanced vision and text
• gpt-4o - Flagship multimodal

​

Switching Models

/model o3

​

Features

​

Reasoning Models

• Explicit chain of thought
• Step-by-step problem solving
• Verification of logic
• Self-correction

​

Vision Capabilities

• Image analysis
• Screenshot understanding
• Diagram interpretation
• UI/UX review

​

Tool Calling

• Function calling
• Parallel tool execution
• Structured outputs
• Multi-step workflows

​

Long Context

• 400K: gpt-5.3-codex
• 200K: o1, o3, gpt-5 series
• 128K: gpt-4 series
• 16K: gpt-3.5-turbo

​

Best Practices

​

Rate Limits

• Requests per minute (RPM)
• Tokens per minute (TPM)
• Tokens per day (TPD)

​

Cost Optimization

• Use gpt-4o-mini for simple tasks
• Use gpt-3.5-turbo for basic operations
• Cache frequently used prompts
• Limit context size when possible

• Use gpt-5.3-codex for complex coding
• Use o3 for difficult reasoning
• Use gpt-5 for maximum capability

​

Token Management

• Monitor token usage in real-time
• Set spending limits in your account
• Use shorter system prompts
• Summarize long conversations

​

Troubleshooting

​

Invalid API Key

1. Verify your key starts with sk-
2. Check that it’s active in your OpenAI account
3. Ensure you have sufficient credits/quota
4. Try regenerating the key

​

Rate Limit Errors

1. Check your tier and limits in the dashboard
2. Implement exponential backoff retry logic
3. Spread requests over time
4. Consider upgrading your tier

​

Insufficient Quota

1. Add credits to your account
2. Check your billing settings
3. Set up auto-recharge if needed
4. Monitor usage to avoid surprises

​

Context Length Exceeded

1. Reduce conversation history
2. Summarize previous context
3. Switch to a model with larger context
4. Split work into smaller chunks

​

Deprecated: Environment Variable Setup

# .env
OPENAI_API_KEY=sk-your-api-key

# forge.yaml
model: gpt-4o

​

Next Steps

• Explore OpenAI’s documentation
• Learn about prompt engineering
• Set up custom commands for your workflow
• Configure temperature and other parameters