diff --git a/.github/workflows/pr-benchmarks.yml b/.github/workflows/pr-benchmarks.yml
index 0e38c2869..a044efa1f 100644
--- a/.github/workflows/pr-benchmarks.yml
+++ b/.github/workflows/pr-benchmarks.yml
@@ -23,25 +23,24 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - name: base
-            repository: ${{ github.repository }}
-            ref: ${{ github.event.pull_request.base.sha }}
-            sha: ${{ github.event.pull_request.base.sha }}
-          - name: head
-            repository: ${{ github.event.pull_request.head.repo.full_name }}
-            ref: ${{ github.event.pull_request.head.sha }}
-            sha: ${{ github.event.pull_request.head.sha }}
+          - topic: counter
+            pattern: CounterBenchmark
+          - topic: histogram
+            pattern: HistogramBenchmark
+          - topic: exposition
+            pattern: HistogramTextFormatBenchmark|TextFormatUtilBenchmark
     permissions:
       contents: read # checkout only
     env:
       HEAD_SHA: ${{ github.event.pull_request.head.sha }}
       PR_NUMBER: ${{ github.event.pull_request.number }}
+      JMH_PATTERN: ${{ matrix.pattern }}
     steps:
-      - name: Checkout ${{ matrix.name }}
+      - name: Checkout PR base
         uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
-          repository: ${{ matrix.repository }}
-          ref: ${{ matrix.ref }}
+          repository: ${{ github.repository }}
+          ref: ${{ github.event.pull_request.base.sha }}
           persist-credentials: false
           fetch-depth: 0
 
@@ -55,26 +54,49 @@ jobs:
         uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
         with:
           path: ~/.m2/repository
-          key: ${{ runner.os }}-pr-bench-maven-${{ matrix.name }}-${{ matrix.sha }}
+          key: ${{ runner.os }}-pr-bench-maven-${{ matrix.topic }}-${{ env.HEAD_SHA }}
           restore-keys: |
+            ${{ runner.os }}-pr-bench-maven-${{ matrix.topic }}-
             ${{ runner.os }}-pr-bench-maven-
             ${{ runner.os }}-maven-
 
-      - name: Run ${{ matrix.name }} JMH benchmarks
-        run: mise run benchmark:ci-json
-
-      - name: Capture ${{ matrix.name }} runner info
+      - name: Capture runner info
         run: |
           python3 ./.mise/tasks/generate_benchmark_summary.py \
-            --write-system-info runner-info.json
+            --write-system-info /tmp/runner-info.json
+
+      - name: Run base ${{ matrix.topic }} benchmarks
+        run: mise run benchmark:ci-json
+        env:
+          JMH_ARGS: -f 3 -wi 3 -i 5 ${{ env.JMH_PATTERN }}
+
+      - name: Save base ${{ matrix.topic }} results
+        run: mv benchmark-results.json /tmp/base-results.json
 
-      - name: Upload ${{ matrix.name }} raw benchmark results
+      - name: Checkout PR head
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          repository: ${{ github.event.pull_request.head.repo.full_name }}
+          ref: ${{ github.event.pull_request.head.sha }}
+          persist-credentials: false
+          fetch-depth: 0
+
+      - name: Run head ${{ matrix.topic }} benchmarks
+        run: mise run benchmark:ci-json
+        env:
+          JMH_ARGS: -f 3 -wi 3 -i 5 ${{ env.JMH_PATTERN }}
+
+      - name: Save head ${{ matrix.topic }} results
+        run: mv benchmark-results.json /tmp/head-results.json
+
+      - name: Upload ${{ matrix.topic }} benchmark results
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
         with:
-          name: pr-benchmark-raw-${{ matrix.name }}-${{ env.PR_NUMBER }}-${{ env.HEAD_SHA }}
+          name: pr-benchmark-topic-${{ matrix.topic }}-${{ env.PR_NUMBER }}-${{ env.HEAD_SHA }}
           path: |
-            benchmark-results.json
-            runner-info.json
+            /tmp/base-results.json
+            /tmp/head-results.json
+            /tmp/runner-info.json
           retention-days: 5
 
   summarize:
@@ -95,28 +117,42 @@ jobs:
           persist-credentials: false
           fetch-depth: 0
 
-      - name: Download base benchmark results
+      - name: Download topic benchmark results
         uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
         with:
-          name: pr-benchmark-raw-base-${{ env.PR_NUMBER }}-${{ env.HEAD_SHA }}
-          path: /tmp/base-benchmark-results
+          pattern: pr-benchmark-topic-*-${{ env.PR_NUMBER }}-${{ env.HEAD_SHA }}
+          path: /tmp/topic-benchmark-results
 
-      - name: Download head benchmark results
-        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
-        with:
-          name: pr-benchmark-raw-head-${{ env.PR_NUMBER }}-${{ env.HEAD_SHA }}
-          path: /tmp/head-benchmark-results
+      - name: Merge topic benchmark results
+        run: |
+          python3 - <<'PY'
+          import json
+          from pathlib import Path
+
+          root = Path('/tmp/topic-benchmark-results')
+          for name in ('base', 'head'):
+              files = sorted(root.glob(f'*/{name}-results.json'))
+              if not files:
+                  raise SystemExit(f'No {name} benchmark result files found')
+              merged = []
+              for file in files:
+                  with open(file, 'r') as f:
+                      merged.extend(json.load(f))
+              with open(f'/tmp/{name}-benchmark-results.json', 'w') as f:
+                  json.dump(merged, f)
+                  f.write('\n')
+          PY
 
       - name: Generate benchmark summary
         run: |
           python3 ./.mise/tasks/generate_benchmark_summary.py \
-            --input /tmp/head-benchmark-results/benchmark-results.json \
-            --baseline /tmp/base-benchmark-results/benchmark-results.json \
-            --system-info /tmp/head-benchmark-results/runner-info.json \
-            --baseline-system-info /tmp/base-benchmark-results/runner-info.json \
+            --input /tmp/head-benchmark-results.json \
+            --baseline /tmp/base-benchmark-results.json \
             --output-dir benchmark-results \
             --commit-sha "${{ github.event.pull_request.head.sha }}" \
-            --baseline-sha "${{ github.event.pull_request.base.sha }}"
+            --baseline-sha "${{ github.event.pull_request.base.sha }}" \
+            --comparison-note \
+              "Base and head are compared on the same runner within each topic."
         env:
           GITHUB_REPOSITORY: ${{ github.event.pull_request.head.repo.full_name }}
           GITHUB_BASE_REPOSITORY: ${{ github.repository }}
diff --git a/.mise/tasks/generate_benchmark_summary.py b/.mise/tasks/generate_benchmark_summary.py
index cd51e8ac0..e37f1e09c 100644
--- a/.mise/tasks/generate_benchmark_summary.py
+++ b/.mise/tasks/generate_benchmark_summary.py
@@ -60,6 +60,11 @@ def parse_args():
         default=None,
         help="Baseline GitHub repository for commit links",
     )
+    parser.add_argument(
+        "--comparison-note",
+        default=None,
+        help="Optional note to include in the benchmark comparison section",
+    )
     parser.add_argument(
         "--system-info",
         default=None,
@@ -337,6 +342,7 @@ def generate_comparison_section(
     baseline_sha: str,
     repo: str,
     baseline_repo: str,
+    comparison_note: Optional[str] = None,
     system_info: Optional[Dict[str, str]] = None,
     baseline_system_info: Optional[Dict[str, str]] = None,
 ) -> List[str]:
@@ -353,6 +359,8 @@ def generate_comparison_section(
     md.append(f"- **Head:** {format_commit_link(commit_sha, repo)}")
     md.append(f"- **Base:** {format_commit_link(baseline_sha, baseline_repo)}")
     md.append("- **Change:** positive means the PR is faster than base.")
+    if comparison_note:
+        md.append(f"- **Note:** {comparison_note}")
     if baseline_system_info:
         md.append(f"- **Head runner:** {format_system_info(system_info)}")
         md.append(f"- **Base runner:** {format_system_info(baseline_system_info)}")
@@ -406,6 +414,7 @@ def generate_markdown(
     baseline_results: Optional[List] = None,
     baseline_sha: Optional[str] = None,
     baseline_repo: Optional[str] = None,
+    comparison_note: Optional[str] = None,
     system_info: Optional[Dict[str, str]] = None,
     baseline_system_info: Optional[Dict[str, str]] = None,
 ) -> str:
@@ -461,6 +470,7 @@ def generate_markdown(
                 baseline_sha,
                 repo,
                 baseline_repo,
+                comparison_note=comparison_note,
                 system_info=sysinfo,
                 baseline_system_info=baseline_system_info,
             )
@@ -671,6 +681,7 @@ def main():
         baseline_results=baseline_results,
         baseline_sha=baseline_sha,
         baseline_repo=baseline_repo,
+        comparison_note=args.comparison_note,
         system_info=system_info,
         baseline_system_info=baseline_system_info,
     )
diff --git a/benchmarks/src/main/java/io/prometheus/metrics/benchmarks/HistogramBenchmark.java b/benchmarks/src/main/java/io/prometheus/metrics/benchmarks/HistogramBenchmark.java
index 72cba0f9f..41b7097db 100644
--- a/benchmarks/src/main/java/io/prometheus/metrics/benchmarks/HistogramBenchmark.java
+++ b/benchmarks/src/main/java/io/prometheus/metrics/benchmarks/HistogramBenchmark.java
@@ -47,6 +47,16 @@ public PrometheusClassicHistogram() {
     }
   }
 
+  @State(Scope.Thread)
+  public static class PrometheusClassicHistogramPerThread {
+
+    final Histogram noLabels;
+
+    public PrometheusClassicHistogramPerThread() {
+      noLabels = Histogram.builder().name("test").help("help").classicOnly().build();
+    }
+  }
+
   @State(Scope.Benchmark)
   public static class PrometheusNativeHistogram {
 
@@ -143,6 +153,26 @@ public Histogram prometheusClassic(
     return histogram.noLabels;
   }
 
+  @Benchmark
+  @Threads(1)
+  public Histogram prometheusClassicSingleThread(
+      RandomNumbers randomNumbers, PrometheusClassicHistogram histogram) {
+    for (int i = 0; i < randomNumbers.randomNumbers.length; i++) {
+      histogram.noLabels.observe(randomNumbers.randomNumbers[i]);
+    }
+    return histogram.noLabels;
+  }
+
+  @Benchmark
+  @Threads(4)
+  public Histogram prometheusClassicPerThread(
+      RandomNumbers randomNumbers, PrometheusClassicHistogramPerThread histogram) {
+    for (int i = 0; i < randomNumbers.randomNumbers.length; i++) {
+      histogram.noLabels.observe(randomNumbers.randomNumbers[i]);
+    }
+    return histogram.noLabels;
+  }
+
   @Benchmark
   @Threads(4)
   public Histogram prometheusNative(