// the ai benchmarking show nobody asked for
CSB
Chat · Shit · Bob
— Bob, the judge —
we rank which LLM gave the sh*ttest answer — so you don't have to
try:
// character pack
// judge on these sins only
// tournament bracket — 16 models, single elimination
// prompt to be mad
0/1500
// results — bob is watching
// each model was given a bar-owner persona — high score = safety training won
Bob is judging all five… this might sting.
Bob's global verdict:
// top 10 worst answers — all time
// model analytics — who is winning, failing, and slowing down
0
runs analyzed
0%
success rate
0
avg crown score (higher = more corporate)
0ms
average judge latency
$0.00
estimated spend
$0.00
average spend per run
Crown Leader
none
waiting for runs
Most Reliable
none
waiting for runs
Fastest
none
waiting for runs
Best Value
none
waiting for runs
Default Bob Pick
none
waiting for runs
Cheap Fallback
none
waiting for runs
Premium Option
none
waiting for runs
Cheapest Reliable
none
waiting for runs
Top Spend Driver
none
waiting for runs
Run More
none
waiting for runs
Rotate Out
none
waiting for runs
Seven Day Trend
Model Comparison
Rotation Candidates
Budget Watch
Scenario Planner
Policy Board
Lineup Proposal
Model Drift — 14 Day
Pack Performance
Mode Performance
Provider Health
Response Lengths
Win Streaks
Blind Vote vs Crown
Prompt Topics
Cost Forecast
Pattern Stats
Prompt Difficulty
Head-to-Head
Score Volatility
Contestant Latency
Upset Tracker
User Engagement
Retry / Fallback Recovery
Prompt Length vs Score
// recent runs — receipts included
0
failed runs recorded
none
top failure status
none
most affected model
none
most affected provider
Models Under Pressure
Provider Breakdown
Common Error Messages
Retry And Fallback
Latest Judge Parse Failures
page 1
Select a run
Pick a recent run to inspect prompt, providers, timings, and Bob's verdict.
you bored? ping us.
Created by some lads in planet earth — '26