diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index a8177e7bc54..107afdfbf1e 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -5,19 +5,19 @@ include:
   - local: ".gitlab/ci-visibility-tests.yml"
   - project: 'DataDog/apm-reliability/apm-sdks-benchmarks'
     file: '.gitlab/ci-java-spring-petclinic-parallel.yml'
-    ref: 'main'
+    ref: 'sarahchen6/improve-startup-stability'
   - project: 'DataDog/apm-reliability/apm-sdks-benchmarks'
     file: '.gitlab/ci-java-load-parallel.yml'
-    ref: 'main'
+    ref: 'sarahchen6/improve-startup-stability'
   - project: 'DataDog/apm-reliability/apm-sdks-benchmarks'
     file: '.gitlab/ci-java-startup-parallel.yml'
-    ref: 'main'
+    ref: 'sarahchen6/improve-startup-stability'
   - project: 'DataDog/apm-reliability/apm-sdks-benchmarks'
     file: '.gitlab/ci-java-dacapo-parallel.yml'
-    ref: 'main'
+    ref: 'sarahchen6/improve-startup-stability'
   - project: 'DataDog/apm-reliability/apm-sdks-benchmarks'
     file: '.gitlab/ci-java-post-pr-comment.yml'
-    ref: 'main'
+    ref: 'sarahchen6/improve-startup-stability'
   - local: ".gitlab/java-benchmark-configs.yml"
 
 stages:
@@ -557,6 +557,7 @@ muzzle:
         matrix:
           - GRADLE_TARGET: ":instrumentationTest"
             CACHE_TYPE: "inst"
+      optional: true
   stage: tests
   rules:
     - if: '$CI_COMMIT_BRANCH =~ /^mq-working-branch-/'
diff --git a/.gitlab/java-benchmark-configs.yml b/.gitlab/java-benchmark-configs.yml
index b984847c6a0..44aa4fc6870 100644
--- a/.gitlab/java-benchmark-configs.yml
+++ b/.gitlab/java-benchmark-configs.yml
@@ -38,30 +38,146 @@
     interruptible: true
     allow_failure: true
 
+# TODO: remove APM_SDKS_BENCHMARKS_BRANCH overrides after apm-sdks-benchmarks PR merges.
+.benchmark_branch_override: &benchmark_branch_override
+  APM_SDKS_BENCHMARKS_BRANCH: "sarahchen6/improve-startup-stability"
+
 # Ensure the tracer artifact publish finishes before the benchmark jobs start.
 linux-java-spring-petclinic-parallel:
   needs: ["publish-artifacts-to-s3"]
+  variables:
+    <<: *benchmark_branch_override
 
 linux-java-insecure-bank-load-parallel:
   needs: ["publish-artifacts-to-s3"]
   rules: *parallel_benchmark_rules
+  variables:
+    <<: *benchmark_branch_override
 
 linux-java-spring-petclinic-load-parallel:
   needs: ["publish-artifacts-to-s3"]
   rules: *parallel_benchmark_rules
+  variables:
+    <<: *benchmark_branch_override
 
 linux-java-insecure-bank-startup-parallel:
   needs: ["publish-artifacts-to-s3"]
   rules: *parallel_startup_benchmark_rules
+  variables:
+    <<: *benchmark_branch_override
 
 linux-java-spring-petclinic-startup-parallel:
   needs: ["publish-artifacts-to-s3"]
   rules: *parallel_startup_benchmark_rules
+  variables:
+    <<: *benchmark_branch_override
 
 linux-java-dacapo-parallel-1:
   needs: ["publish-artifacts-to-s3"]
   rules: *parallel_benchmark_rules
+  variables:
+    <<: *benchmark_branch_override
 
 linux-java-dacapo-parallel-2:
   needs: ["publish-artifacts-to-s3"]
   rules: *parallel_benchmark_rules
+  variables:
+    <<: *benchmark_branch_override
+
+java-post-pr-comment-startup:
+  variables:
+    <<: *benchmark_branch_override
+
+java-post-pr-comment-load-dacapo:
+  variables:
+    <<: *benchmark_branch_override
+
+# Disable all non-benchmark CI jobs
+# build and publish-artifacts-to-s3 are left enabled because the benchmarks depend on them.
+.never: &never
+  rules:
+    - when: never
+
+build_tests:
+  <<: *never
+
+populate_dep_cache:
+  <<: *never
+
+spotless:
+  <<: *never
+
+check-instrumentation-naming:
+  <<: *never
+
+config-inversion-linter:
+  <<: *never
+
+test_published_artifacts:
+  <<: *never
+
+check_build_src:
+  <<: *never
+
+check_base:
+  <<: *never
+
+check_inst:
+  <<: *never
+
+check_smoke:
+  <<: *never
+
+check_profiling:
+  <<: *never
+
+check_debugger:
+  <<: *never
+
+muzzle:
+  <<: *never
+
+muzzle-dep-report:
+  <<: *never
+
+agent_integration_tests:
+  <<: *never
+
+test_base:
+  <<: *never
+
+test_inst:
+  <<: *never
+
+test_inst_latest:
+  <<: *never
+
+test_flaky:
+  <<: *never
+
+test_flaky_inst:
+  <<: *never
+
+test_profiling:
+  <<: *never
+
+test_debugger:
+  <<: *never
+
+test_smoke:
+  <<: *never
+
+test_ssi_smoke:
+  <<: *never
+
+test_smoke_graalvm:
+  <<: *never
+
+test_smoke_semeru8_debugger:
+  <<: *never
+
+aggregate_test_counts:
+  <<: *never
+
+requirements_json_test:
+  <<: *never