Client configuration

Stratix / AsyncStratix client — construction, env vars, timeout, retries, per-request overrides.

Sync client

from layerlens import Stratix

client = Stratix() # picks up LAYERLENS_STRATIX_API_KEY
client = Stratix(api_key="your-key") # explicit

Async client

from layerlens import AsyncStratix
client = AsyncStratix()

Aliases

For backward compatibility:

from layerlens import Stratix, AsyncStratix # primary
from layerlens import Client, AsyncClient # aliases
from layerlens import Atlas, AsyncAtlas # legacy aliases
from layerlens import PublicClient, AsyncPublicClient

Constructor parameters

Stratix(api_key, base_url, timeout, max_retries) and AsyncStratix(...):

Parameter	Type	Required	Default	Description
api_key	str | None	yes*	None	Stratix API key
base_url	str | httpx.URL | None	no	Stratix API URL	Custom base URL
timeout	float | httpx.Timeout | None	no	10 minutes (600s)	Timeout config
max_retries	int	no	2	Max retries (429, 5xx)

*Required unless set via env var.

Environment variables

Variable	Default	Notes
LAYERLENS_STRATIX_API_KEY	(required)	Primary key env var
LAYERLENS_STRATIX_BASE_URL	https://api.layerlens.ai/api/v1	Override for self-hosted / staging
LAYERLENS_ATLAS_API_KEY	—	Legacy alias
LAYERLENS_ATLAS_BASE_URL	—	Legacy alias

Timeout

client = Stratix(timeout=30.0) # 30 seconds for all requests
client = Stratix(timeout=10*60) # 10 minutes (default)

Per-request override

.with_options() returns a copy of the client with overridden options:

result = client.with_options(timeout=300.0).evaluations.create(model=..., benchmark=...)

Retries

The client retries on 429, 500, 502, 503, 504 with exponential backoff and respects Retry-After:

client = Stratix(max_retries=5) # for batch-heavy workloads
client = Stratix(max_retries=0) # disable

See also

• Errors — exception hierarchy
• Authentication
• Public client