OpenAI Setup

Connect Browser Operator directly to OpenAI’s API for access to GPT-5, GPT-4.1, and their variants.

Prerequisites

• An OpenAI account at platform.openai.com
• Valid payment method added to your OpenAI account
• Active API credits or billing enabled

Getting Your API Key

1. Visit OpenAI API Keys
2. Click “Create new secret key”
3. Choose your Project (recommended: create a dedicated “Browser Operator” project)
4. Give your key a descriptive name (e.g., “Browser Operator”)
5. Copy the key — project keys start with sk-proj-... (older org/user keys may start with sk-...)

⚠️ Important: Save your API key immediately. You won’t be able to see it again after closing the dialog.

Configuring Browser Operator

1. Open Browser Operator
2. Click the AI Chat button in the top toolbar
3. Click the Settings icon (⚙️) in the AI Chat panel
4. Select “OpenAI” from the Provider dropdown
5. Paste your API key in the API Key field
6. Select your preferred model (recommended: gpt-5-mini for best balance, gpt-5-nano for speed/cost, or gpt-5 for maximum quality)
7. Click “Save”

Available Models (October 2025)

GPT-5 Series (Released August 2025)

GPT-4.1 Series (Released April 2025)

Notes:

• Pricing shown as Input / Output tokens
• GPT-5 achieves 94.6% on AIME 2025 math, 74.9% on SWE-bench coding
• GPT-5 responses are ~45% less likely to contain factual errors than GPT-4o
• GPT-5 models support 4 reasoning levels: minimal, low, medium, high
• GPT-4.1 excels at coding benchmarks and long-context understanding (1M tokens)
• Prompt caching available with 75% discount on cached tokens
• Batch API available for 50% discount on asynchronous processing
• Always verify current rates at OpenAI Pricing

Model Selection Guide

Choose gpt-5 when:

• You need the absolute best performance available
• State-of-the-art coding, math, and reasoning
• Factual accuracy is critical (45% fewer errors)
• Expert-level responses required
• Can adjust reasoning levels (minimal/low/medium/high) for speed vs quality

Choose gpt-5-mini when:

• You want excellent performance at lower cost
• Best balance of quality, speed, and price
• General purpose assistant and coding tasks
• Most popular choice for production workloads

Choose gpt-5-nano when:

• You need ultra-fast responses at very low cost
• High-volume workloads and simple tasks
• Great default for everyday chatting in Browser Operator
• Best cost-to-performance ratio

Choose gpt-4.1 when:

• You need 1M token input context for very long documents
• Excellent coding and long-context understanding
• Strong general capability at competitive pricing
• 26% cheaper than previous generation

Choose gpt-4.1-mini or gpt-4.1-nano when:

• You want fast responses with 1M token context support
• Cost optimization for high-volume applications
• Good quality at extremely low prices

Managing Usage and Costs

Setting Usage Limits

1. Visit OpenAI Usage Limits
2. Set hard and soft usage limits
3. Configure email notifications
4. Set monthly budget caps

Monitoring Usage

1. Visit OpenAI Usage Dashboard
2. View detailed usage by:
   • Model
   • Date range
   • Cost breakdown
   • API key (if multiple)

Troubleshooting

“Invalid API Key” Error

Solutions:

• Verify you copied the complete key (including sk-proj- or sk- prefix)
• Check that the key hasn’t been revoked in your OpenAI dashboard
• Ensure there are no extra spaces before/after the key
• Try generating a new key

“Insufficient Credits” Error

Solutions:

• Add a payment method at OpenAI Billing
• Check your usage limits aren’t exceeded
• Verify your payment method is valid
• Add prepaid credits to your account

“Rate Limit Exceeded” Error

Solutions:

• Wait a few seconds and retry
• Reduce request frequency
• Upgrade your OpenAI tier for higher limits
• Implement exponential backoff in your requests

“Model Not Found” Error

Solutions:

• Verify the model name is correct (e.g., gpt-5-mini)
• Check if you have access to the model (some require tier upgrades)
• Ensure your API key has permission to use the model
• Try a different model like gpt-5-nano

Slow Response Times

Solutions:

• Use faster models (gpt-5-nano, gpt-4.1-nano) or lower reasoning levels
• Reduce max_tokens parameter
• Simplify your prompts
• Check OpenAI status at status.openai.com

OpenAI Tier System

OpenAI uses a tier system that affects rate limits and access:

Higher tiers unlock:

• Increased rate limits
• Access to newer models
• Higher throughput
• Priority support

Advanced Configuration

Organization Settings

If you’re part of multiple OpenAI organizations:

1. Visit OpenAI Settings
2. Note your Organization ID
3. Select the appropriate organization when creating API keys

Comparing OpenAI vs OpenRouter

Next Steps

• Getting Started - Complete setup guide
• User Guide - Learn Browser Operator features
• Developer Guide - Build custom agents
• OpenAI Documentation - Learn more about OpenAI API

Support

• OpenAI API Issues: OpenAI Help Center
• Browser Operator Issues: GitHub Issues
• Community: Browser Operator Discord