Quarterly rebench

Recipe — quarterly model re-evaluation as the frontier moves.

Every quarter, re-run the model evaluation with the current candidate set. The frontier moves; what won last quarter may not win this one.

from layerlens import Stratix
client = Stratix()

# Pull current top reasoning models (filter to your shape)
public_models = client.public.models.get(
 categories=["transformer"],
 sort_by="releasedAt",
 order="desc",
 page=1,
 page_size=10,
)

benchmark = client.benchmarks.get_by_key("your-team-benchmark")

# Run an evaluation per candidate
for model in public_models.models:
 evaluation = client.evaluations.create(model=model, benchmark=benchmark)
 print(f"Started {evaluation.id} for {model.key}")

After all complete, compare via samples/core/compare_evaluations.py.

See also

• Quarterly reports
• Workflow: Compare