OpenAIGPT-4o
OpenAI's workhorse model. Fast (15.6s avg) and token-efficient but lower checkout completion than frontier models. Moderate tool error rate (8.4%). Never needs nudging — either completes or gives up cleanly. Good for high-volume, cost-sensitive benchmarking where speed matters more than checkout depth.
- Avg tokens per run
- 34,703
- Avg duration
- 16.5s
- Turns to checkout
- 8.3
Scores come from runs on UCP stores, not estimates: every figure is a tool call, a checkout attempt or a store’s answer.
Shopping score breakdown
- Checkout rate40%14.7%
- Cart rate20%35.9%
- Search rate10%72.2%
- Turn efficiency15%46%
- Token efficiency15%28%
26.9% of runs failed with errors.
Top stores
Development stores are hidden.
| Store | Checkout | Cart |
|---|---|---|
| everlane.com | 61.5% | 61.5% |
| allbirds.com | 33.3% | 33.3% |
| store hidden | 33.3% | 33.3% |
| houseofparfum.nl | 30.8% | 38.5% |
| ucp.travel | 28.6% | 28.6% |
| store hidden | 25% | 50% |
| kyliecosmetics.com | 20% | 20% |
| store hidden | 9.1% | 45.5% |
| store hidden | 8.3% | 91.7% |
| store hidden | 0% | 0% |
Known issues
- MediumLow Checkout Completion
33% checkout rate despite fast execution. Gives up rather than retrying on errors.
- MediumModerate Error Rate
8.4% tool error rate — higher than Claude and Gemini models.
Token usage
- Avg per run
- 34,703
- Average of all models
- 78,415
- Against the average
- −56%
- Median (p50)
- 22,584
- p90
- 85,675
Prompt and completion
Daily average tokens per run, last 30 days
Oct 21Total 24,858Prompt 24,323Completion 535
Range
- Min
- 0
- p25
- 8,180
- p50
- 22,584
- p75
- 39,724
- p90
- 85,675
- Max
- 437,563
Test GPT-4o on your store
Start a run to see how GPT-4o handles your store’s checkout, call by call.
Compare with other models
Rates are checkout rates. Full comparison view