diff --git a/tests/e2e/coverage_registry/rate_limiting.yaml b/tests/e2e/coverage_registry/rate_limiting.yaml
index 2ed1a74953c..c93a53afd3e 100644
--- a/tests/e2e/coverage_registry/rate_limiting.yaml
+++ b/tests/e2e/coverage_registry/rate_limiting.yaml
@@ -1,3 +1,3 @@
 # Rate limiting behaviors. Grounded in litellm/proxy/hooks/parallel_request_limiter.py.
 - {id: rate_limiting.key.tpm.under_limit_allows, module: rate_limiting, tier: P0, scope: key, limit: tpm, behavior: allows_under_limit, assertions: [persists_limit, allows_request], source: "parallel_request_limiter.py:145-260", rationale: "A virtual key below its TPM quota can call the data plane"}
-- {id: rate_limiting.key.tpm.over_limit_blocks, module: rate_limiting, tier: P0, scope: key, limit: tpm, behavior: blocks_over_limit, assertions: [persists_limit, records_usage, blocks_next_request], source: "parallel_request_limiter.py:145-260", rationale: "A virtual key above its TPM quota is rejected with 429"}
+- {id: rate_limiting.key.tpm.over_limit_blocks, module: rate_limiting, tier: P0, scope: key, limit: tpm, behavior: blocks_over_limit, assertions: [persists_limit, returns_usage, blocks_next_request], source: "parallel_request_limiter.py:145-260", rationale: "A virtual key above its TPM quota is rejected with 429"}
diff --git a/tests/e2e/rate_limiting/conftest.py b/tests/e2e/rate_limiting/conftest.py
index 1bbd2374ac0..e374b35e460 100644
--- a/tests/e2e/rate_limiting/conftest.py
+++ b/tests/e2e/rate_limiting/conftest.py
@@ -5,6 +5,6 @@ import pytest
 from rate_limiting_client import RateLimitingClient, build_client
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def client() -> RateLimitingClient:
     return build_client()