This site is in development.
Features and figures may change.
Contact us
Skip to content
Sample data. These figures are not a customer bill.
Learn more
Northwind AI /
All workspaces
Workspace
Search pages, models and records
Ctrl K
Pay as you go
Sample user
Pay as you go · Builder view
SU
Home
Routing
Quality tests
All envs
Environment
Last 30 days
Range
Provider routing
Auto router
Router settings
Quality tests
Caching
Prompt trimming
Batch
Activity
Savings
,
2
open recommendations
Models
Chat
More
←
Quality tests
New quality test
Route
Support reply
Ticket triage
Document extract
Eval backfill
Candidate models
GPT-4.1
GPT-4.1 mini
Claude Sonnet 4.5
Claude Haiku 4.5
Gemini 2.5 Pro
Gemini 2.5 Flash
Grok 3
Mistral Large
DeepSeek V3
Llama 3.3 70B
Sample %
Share of the route's traffic the replay uses. Capped at 20%.
Pass threshold
The customer's quality bar, from 0.50 to 1.
Scorer
Task success
Scores the candidate against the route's task tag.
Judge model
A stronger model grades the candidate. The judge's tokens are counted as cost.
Exact match
For structured routes, the candidate must match the schema and the expected fields.
I confirm the estimated replay cost. The test stops at the replay cap you set.
Queue shadow test
