diff --git a/.github/workflows/hyperfine_prove.yaml b/.github/workflows/hyperfine_prove.yaml
new file mode 100644
index 000000000..e2fd3e8e2
--- /dev/null
+++ b/.github/workflows/hyperfine_prove.yaml
@@ -0,0 +1,247 @@
+name: Hyperfine Prove Benchmark
+
+on:
+  workflow_dispatch:
+    inputs:
+      program:
+        description: 'Bench program to prove (e.g. all_instructions_64, bench_32k)'
+        required: false
+        default: 'all_instructions_64'
+      runs:
+        description: 'Number of hyperfine runs'
+        required: false
+        default: '3'
+  pull_request:
+    types: [labeled]
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+env:
+  CARGO_TERM_COLOR: always
+  BENCH_PROGRAM: ${{ github.event.inputs.program || 'all_instructions_64' }}
+  BENCH_RUNS: ${{ github.event.inputs.runs || '3' }}
+
+jobs:
+  build-program:
+    name: Build bench program
+    if: github.event_name == 'workflow_dispatch' || github.event.label.name == 'bench-prove'
+    runs-on: ubuntu-24.04
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Fetch from cache
+        uses: actions/cache@v4
+        id: cache
+        with:
+          path: bench_programs/*.elf
+          key: prove-programs-${{ hashFiles('executor/programs/asm/*') }}
+
+      - name: Setup Rust Environment
+        if: steps.cache.outputs.cache-hit != 'true'
+        uses: ./.github/actions/setup-rust
+
+      - name: Build programs
+        if: steps.cache.outputs.cache-hit != 'true'
+        run: |
+          mkdir -p bench_programs
+          make -j compile-programs-asm
+          cp executor/program_artifacts/asm/*.elf bench_programs/
+
+      - name: Upload programs
+        uses: actions/upload-artifact@v4
+        with:
+          name: bench-programs
+          path: bench_programs/*.elf
+          retention-days: 1
+
+  build-base-binary:
+    name: Build CLI for base
+    if: github.event.label.name == 'bench-prove'
+    runs-on: ubuntu-24.04
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ github.event.pull_request.base.sha }}
+
+      - name: Setup Rust Environment
+        uses: ./.github/actions/setup-rust
+
+      - name: Fetch Rust cache
+        uses: Swatinem/rust-cache@v2
+        with:
+          shared-key: "lambda-vm"
+
+      - name: Build binary
+        run: |
+          cargo build --release --bin cli
+          mkdir -p bench-bin
+          cp target/release/cli bench-bin/cli-base
+
+      - name: Upload binary
+        uses: actions/upload-artifact@v4
+        with:
+          name: cli-base
+          path: bench-bin/cli-base
+          retention-days: 1
+
+  build-head-binary:
+    name: Build CLI for head
+    if: github.event_name == 'workflow_dispatch' || github.event.label.name == 'bench-prove'
+    runs-on: ubuntu-24.04
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ github.event.pull_request.head.sha || github.sha }}
+
+      - name: Setup Rust Environment
+        uses: ./.github/actions/setup-rust
+
+      - name: Fetch Rust cache
+        uses: Swatinem/rust-cache@v2
+        with:
+          shared-key: "lambda-vm"
+
+      - name: Build binary
+        run: |
+          cargo build --release --bin cli
+          mkdir -p bench-bin
+          cp target/release/cli bench-bin/cli-head
+
+      - name: Upload binary
+        uses: actions/upload-artifact@v4
+        with:
+          name: cli-head
+          path: bench-bin/cli-head
+          retention-days: 1
+
+  run-prove-benchmark:
+    name: Run prove benchmark
+    needs: [build-program, build-base-binary, build-head-binary]
+    if: always() && needs.build-program.result == 'success' && needs.build-head-binary.result == 'success' && needs.build-base-binary.result != 'failure'
+    runs-on: ubuntu-24.04
+    timeout-minutes: 120
+    steps:
+      - name: Install Hyperfine
+        uses: taiki-e/install-action@v2
+        with:
+          tool: hyperfine@1.19
+
+      - name: Download head binary
+        uses: actions/download-artifact@v4
+        with:
+          name: cli-head
+          path: bench-bin
+
+      - name: Download base binary
+        if: github.event.label.name == 'bench-prove'
+        uses: actions/download-artifact@v4
+        with:
+          name: cli-base
+          path: bench-bin
+
+      - name: Download bench programs
+        uses: actions/download-artifact@v4
+        with:
+          name: bench-programs
+          path: bench_programs
+
+      - name: Benchmark prove
+        id: run-benchmark
+        run: |
+          chmod +x bench-bin/cli-*
+
+          PROGRAM="${{ env.BENCH_PROGRAM }}"
+          ELF="bench_programs/${PROGRAM}.elf"
+
+          if [ ! -f "$ELF" ]; then
+            echo "::error::Program $PROGRAM not found"
+            exit 1
+          fi
+
+          PROOF_HEAD="/tmp/proof_head.cbor"
+
+          if [ -f bench-bin/cli-base ]; then
+            PROOF_BASE="/tmp/proof_base.cbor"
+            hyperfine \
+              --warmup 0 \
+              --runs "${{ env.BENCH_RUNS }}" \
+              --prepare "rm -f $PROOF_BASE $PROOF_HEAD" \
+              -n "base prove $PROGRAM" \
+              "./bench-bin/cli-base prove $ELF --output $PROOF_BASE" \
+              -n "head prove $PROGRAM" \
+              "./bench-bin/cli-head prove $ELF --output $PROOF_HEAD" \
+              --export-markdown results.md \
+              --export-json results.json
+          else
+            hyperfine \
+              --warmup 0 \
+              --runs "${{ env.BENCH_RUNS }}" \
+              --prepare "rm -f $PROOF_HEAD" \
+              -n "prove $PROGRAM" \
+              "./bench-bin/cli-head prove $ELF --output $PROOF_HEAD" \
+              --export-markdown results.md
+          fi
+
+          echo "has_results=true" >> "$GITHUB_OUTPUT"
+
+      - name: Detect regression
+        id: detect-regression
+        if: steps.run-benchmark.outputs.has_results == 'true'
+        run: |
+          if [ ! -f results.json ]; then
+            echo "No base vs head comparison (dispatch-only run), skipping regression check"
+            exit 0
+          fi
+
+          BASE_MEAN=$(jq '.results[0].mean' results.json)
+          HEAD_MEAN=$(jq '.results[1].mean' results.json)
+
+          # Flag regression if head is >5% slower than base
+          REGRESSED=$(awk -v bm="$BASE_MEAN" -v hm="$HEAD_MEAN" 'BEGIN {
+            if (bm <= 0) { print "false"; exit }
+            pct = ((hm - bm) / bm) * 100
+            printf "base=%.2fs  head=%.2fs  change=%+.1f%%\n", bm, hm, pct > "/dev/stderr"
+            print (pct > 5 ? "true" : "false")
+          }')
+
+          echo "regressed=$REGRESSED" >> "$GITHUB_OUTPUT"
+
+      - name: Build comment body
+        if: steps.detect-regression.outputs.regressed == 'true'
+        run: |
+          {
+            echo "Prove Benchmark Results :lock:"
+            echo ""
+            echo "Program: \`${{ env.BENCH_PROGRAM }}\` | Runs: \`${{ env.BENCH_RUNS }}\`"
+            echo ""
+            cat results.md
+          } > comment_body.md
+
+      - name: Find comment
+        if: steps.detect-regression.outputs.regressed == 'true' && github.event.pull_request.number
+        uses: peter-evans/find-comment@v3
+        id: fc
+        with:
+          issue-number: ${{ github.event.pull_request.number }}
+          comment-author: "github-actions[bot]"
+          body-includes: Prove Benchmark Results
+
+      - name: Create comment
+        if: steps.fc.outputs.comment-id == '' && steps.detect-regression.outputs.regressed == 'true' && github.event.pull_request.number
+        uses: peter-evans/create-or-update-comment@v4
+        with:
+          issue-number: ${{ github.event.pull_request.number }}
+          body-path: comment_body.md
+
+      - name: Update comment
+        if: steps.fc.outputs.comment-id != '' && steps.detect-regression.outputs.regressed == 'true' && github.event.pull_request.number
+        uses: peter-evans/create-or-update-comment@v4
+        with:
+          comment-id: ${{ steps.fc.outputs.comment-id }}
+          body-path: comment_body.md
+          edit-mode: replace
diff --git a/Makefile b/Makefile
index c02bffc49..7627a06e1 100644
--- a/Makefile
+++ b/Makefile
@@ -1,7 +1,7 @@
 .PHONY: deps deps-linux deps-macos prepare-test-data compile-programs-asm compile-programs-rust compile-bench \
 compile-programs clean-asm clean-rust clean-bench clean-shared clean test test-asm test-no-compile \
 test-asm-no-compile test-rust test-rust-no-compile test-executor flamegraph-prover \
-test-fast test-prover test-prover-all build check clippy fmt lint
+test-fast test-prover test-prover-all build check bench-prove clippy fmt lint
 
 UNAME := $(shell uname)
 
@@ -174,6 +174,10 @@ build:
 check:
 	cargo check --workspace
 
+bench-prove: compile-programs-asm
+	cargo build --release -p cli
+	./scripts/bench_prove.sh $(BENCH_PROVE_PROGRAM)
+
 # === Linting ===
 # op_ref: We pass big integers (U256/U384) and field elements by reference since operator
 # impls delegate to &self internally, avoiding unnecessary 32-48 byte copies.
diff --git a/scripts/bench_prove.sh b/scripts/bench_prove.sh
new file mode 100755
index 000000000..37749c3cb
--- /dev/null
+++ b/scripts/bench_prove.sh
@@ -0,0 +1,116 @@
+#!/usr/bin/env bash
+#
+# Prove Benchmark Script
+# Benchmarks the CLI prove command on the current branch using hyperfine.
+#
+# Usage:
+#   ./bench_prove.sh              # Benchmark all bench programs
+#   ./bench_prove.sh <program>    # Benchmark a specific program (e.g. bench_32k)
+#
+# Environment variables:
+#   BENCH_PROVE_RUNS      Number of hyperfine runs (default: 3)
+#   BENCH_PROVE_WARMUP    Number of warmup runs (default: 0)
+#
+# Requires: hyperfine, jq
+#
+
+set -euo pipefail
+
+for cmd in hyperfine jq; do
+    if ! command -v "$cmd" &>/dev/null; then
+        echo "Error: $cmd is required but not installed."
+        exit 1
+    fi
+done
+
+# Colors
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m'
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+ROOT_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
+TMP_DIR="/tmp/bench_prove"
+BENCH_ARTIFACTS_DIR="$ROOT_DIR/executor/program_artifacts/asm"
+
+RUNS="${BENCH_PROVE_RUNS:-3}"
+WARMUP="${BENCH_PROVE_WARMUP:-0}"
+
+if [ "${1:-}" = "-h" ] || [ "${1:-}" = "--help" ]; then
+    echo "Prove Benchmark Script"
+    echo "Benchmarks the CLI prove command using hyperfine."
+    echo ""
+    echo "Usage:"
+    echo "  ./bench_prove.sh              # Benchmark all bench programs"
+    echo "  ./bench_prove.sh <program>    # Benchmark a specific program (e.g. bench_32k)"
+    echo ""
+    echo "Environment variables:"
+    echo "  BENCH_PROVE_RUNS      Number of runs (default: 3)"
+    echo "  BENCH_PROVE_WARMUP    Number of warmup runs (default: 0)"
+    exit 0
+fi
+
+PROGRAM="${1:-}"
+
+echo -e "${GREEN}=== Prove Benchmark ===${NC}"
+echo -e "${YELLOW}Runs: $RUNS | Warmup: $WARMUP${NC}"
+
+# Find CLI binary
+CLI="$ROOT_DIR/target/release/cli"
+if [ ! -f "$CLI" ]; then
+    echo -e "${RED}Error: CLI binary not found at $CLI. Build with: cargo build --release -p cli${NC}"
+    exit 1
+fi
+
+# Collect ELFs to benchmark
+if [ -n "$PROGRAM" ]; then
+    ELF="$BENCH_ARTIFACTS_DIR/$PROGRAM.elf"
+    if [ ! -f "$ELF" ]; then
+        echo -e "${RED}Error: Program '$PROGRAM' not found at $ELF${NC}"
+        exit 1
+    fi
+    ELFS=("$ELF")
+else
+    shopt -s nullglob
+    ELFS=("$BENCH_ARTIFACTS_DIR"/*.elf)
+    shopt -u nullglob
+    if [ ${#ELFS[@]} -eq 0 ]; then
+        echo -e "${RED}Error: No ELF files found in $BENCH_ARTIFACTS_DIR. Run: make compile-programs-asm${NC}"
+        exit 1
+    fi
+fi
+
+# Setup output directory
+mkdir -p "$TMP_DIR"
+
+# Run benchmarks
+for elf in "${ELFS[@]}"; do
+    name=$(basename "$elf" .elf)
+    proof_file="$TMP_DIR/${name}_proof.cbor"
+    echo -e "${YELLOW}--- $name ---${NC}"
+    hyperfine \
+        --warmup "$WARMUP" \
+        --runs "$RUNS" \
+        --prepare "rm -f '$proof_file'" \
+        -n "prove $name" \
+        "'$CLI' prove '$elf' --output '$proof_file'" \
+        --export-markdown "$TMP_DIR/$name.md" \
+        --export-json "$TMP_DIR/$name.json"
+done
+
+# Summary
+echo ""
+echo -e "${GREEN}=== Results ===${NC}"
+for elf in "${ELFS[@]}"; do
+    name=$(basename "$elf" .elf)
+    if [ -f "$TMP_DIR/$name.json" ]; then
+        mean=$(jq -r '.results[0].mean' "$TMP_DIR/$name.json")
+        stddev=$(jq -r '.results[0].stddev' "$TMP_DIR/$name.json")
+        printf "%-20s  mean: %8.2fs  stddev: %6.2fs\n" "$name" "$mean" "$stddev"
+    fi
+done
+
+echo ""
+echo -e "${GREEN}Markdown reports: $TMP_DIR/*.md${NC}"
+echo -e "${GREEN}JSON results:     $TMP_DIR/*.json${NC}"