Traffic and cost, by tier
Every request the gateway has logged (Postgres, or this process's memory in demo mode), with its estimated cost against GPT-4o list prices.
connecting
provider: …
Totals
- est. saving vs gpt-4o
- - send prompts to see it
- est. saved
- - at list prices
- requests
- - - from cache
- avg latency
- - est. cost -
Lanes
0 requestssimplescore 0-1
-
Factual Q&A, definitions, quick lookups
0 requests0%
mediumscore 2-3
-
Explanations, comparisons, writing
0 requests0%
complexscore 4+
-
Code, algorithms, multi-step reasoning
0 requests0%
Console
requests sent here are logged like any otherChat
POST /v1/chat/completions
tier:
Each message is scored by the rules and sent to the matching lane. Try one:
Session log
this browser tabNo requests yet
Costs use list prices and, in demo mode, estimated token counts. They show the pricing logic, not measured savings.