OpenAI

OpenAI provides GPT models known for strong general reasoning. It’s the default AI engine in Codemetry.

Setup

Get an API Key

Visit platform.openai.com/api-keys to create an API key.
Set Environment Variable
Terminal window
```
export OPENAI_API_KEY=sk-proj-...
```
Or add to your .env file:
```
OPENAI_API_KEY=sk-proj-...
```

Configure Codemetry

'ai' => [
    'enabled' => true,
    'engine' => 'openai',
    'model' => null,  // Uses gpt-4o-mini by default
],

Run Analysis

php artisan codemetry:analyze --days=7 --ai=1

Available Models

Model	Default	Cost	Quality	Speed
`gpt-4o-mini`	✓	Low	Good	Fast
`gpt-4o`		Medium	Excellent	Medium
`gpt-4-turbo`		High	Excellent	Medium
`gpt-3.5-turbo`		Very Low	Acceptable	Very Fast

Choosing a Model

For routine analysis: Use gpt-4o-mini (default)

Best cost/quality balance
Fast response times
Sufficient for metric explanations

For detailed investigation: Use gpt-4o

Better reasoning about complex patterns
More nuanced explanations
Worth the extra cost for deep dives

# Override model for single run
php artisan codemetry:analyze --days=7 --ai=1 --ai-model=gpt-4o

Or in configuration:

'ai' => [
    'engine' => 'openai',
    'model' => 'gpt-4o',  // Override default
],

API Key Options

Codemetry checks for API keys in this order:

CODEMETRY_AI_API_KEY — Unified key for any engine
OPENAI_API_KEY — OpenAI-specific key

# Option 1: Unified (works if you switch engines)
export CODEMETRY_AI_API_KEY=sk-proj-...

# Option 2: Engine-specific
export OPENAI_API_KEY=sk-proj-...

Cost Management

OpenAI charges per token. Codemetry sends minimal data:

~500-1000 input tokens per day analyzed
~200-500 output tokens per summary

Estimated costs per 30-day analysis:

Model	Input	Output	Total
gpt-4o-mini	~$0.005	~$0.01	~$0.015
gpt-4o	~$0.08	~$0.15	~$0.23
gpt-3.5-turbo	~$0.002	~$0.003	~$0.005

Troubleshooting

“Invalid API key”

Cause: API key is incorrect or expired.

Solution:

Verify key at platform.openai.com/api-keys
Check for typos or extra whitespace
Ensure key has not been revoked

“Rate limit exceeded”

Cause: Too many requests or token usage exceeded tier limits.

Solution:

Wait and retry
Check usage at platform.openai.com/usage
Consider upgrading your OpenAI tier

“Model not found”

Cause: Invalid model name or model not available on your account.

Solution:

Check model name spelling
Verify model availability for your account tier
Use a different model

AI Summary Missing

Cause: AI request failed, analysis continued without it.

Check:

ai_unavailable confounder in output indicates AI failed
Check Laravel logs for detailed error messages
Verify network connectivity to OpenAI