| Route | p50 latency | p95 latency | Throughput | Success rate | ShareShare of traffic |
|---|---|---|---|---|---|
| OpenAI direct | 2.25 s | 18.42 s | 134 tok/s | 99.3% | 67% |
| InferHub | 7.49 s | 82.59 s | 54 tok/s | 98.5% | 15% |
| Venice AI direct | 4.11 s | 10.11 s | 99 tok/s | 100.0% | 8% |
| OpenRouter | 3.04 s | 10.11 s | 134 tok/s | 100.0% | 7% |
| Microsoft Azure | 7.49 s | 61.18 s | 73 tok/s | 100.0% | 3% |
| Mordiem | 4.11 s | 18.42 s | - | 99.1% | 0% |
Any OpenAI-compatible client works. Add /min30 to the path to only buy at 30% off or better.
Read the quickstart ↗from openai import OpenAI client = OpenAI( api_key="inf_your_key_here", base_url="$SURPLUS_BASE_URL" ) response = client.chat.completions.create( model="gpt-6-luna", messages=[{"role": "user", "content": "Hello!"}] )
SURPLUS_BASE_URL
https://api.surplusintelligence.ai/v1