diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake2s_constraint.cpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake2s_constraint.cpp
index bc0bcef35c84..82d8457848d0 100644
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake2s_constraint.cpp
+++ b/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake2s_constraint.cpp
@@ -8,7 +8,6 @@
 #include "barretenberg/common/assert.hpp"
 #include "barretenberg/stdlib/hash/blake2s/blake2s.hpp"
 #include "barretenberg/stdlib/primitives/byte_array/byte_array.hpp"
-#include "round.hpp"
 
 namespace acir_format {
 
@@ -24,16 +23,12 @@ template <typename Builder> void create_blake2s_constraints(Builder& builder, co
 
     for (const auto& witness_index_num_bits : constraint.inputs) {
         auto witness_index = witness_index_num_bits.blackbox_input;
-        auto num_bits = witness_index_num_bits.num_bits;
-
-        // XXX: The implementation requires us to truncate the element to the nearest byte and not bit
-        auto num_bytes = round_to_nearest_byte(num_bits);
-        BB_ASSERT_LTE(num_bytes, 32U, "Input num_bytes exceeds 32 per element in blake2s");
-
         field_ct element = to_field_ct(witness_index, builder);
 
-        // byte_array_ct(field, num_bytes) constructor adds range constraints for each byte
-        byte_array_ct element_bytes(element, num_bytes);
+        // byte_array_ct(field, num_bytes) constructor adds range constraints for each byte. Note that num_bytes =
+        // ceil(witness_index_num_bits.num_bits/8). Here, num_bits is set to 8 when constructing the vector of inputs in
+        // the Blake2s constraint. Hence, we set num_bytes = 1.
+        byte_array_ct element_bytes(element, 1);
 
         // Safe write: both arr and element_bytes are constrained
         arr.write(element_bytes);
diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake3_constraint.cpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake3_constraint.cpp
index b36189451ac3..14b0b0f3c321 100644
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake3_constraint.cpp
+++ b/barretenberg/cpp/src/barretenberg/dsl/acir_format/blake3_constraint.cpp
@@ -8,7 +8,6 @@
 #include "barretenberg/common/assert.hpp"
 #include "barretenberg/stdlib/hash/blake3s/blake3s.hpp"
 #include "barretenberg/stdlib/primitives/byte_array/byte_array.hpp"
-#include "round.hpp"
 
 namespace acir_format {
 
@@ -22,15 +21,12 @@ template <typename Builder> void create_blake3_constraints(Builder& builder, con
 
     for (const auto& witness_index_num_bits : constraint.inputs) {
         auto witness_index = witness_index_num_bits.blackbox_input;
-        auto num_bits = witness_index_num_bits.num_bits;
-
-        // XXX: The implementation requires us to truncate the element to the nearest byte and not bit
-        auto num_bytes = round_to_nearest_byte(num_bits);
-        BB_ASSERT_LTE(num_bytes, 32U, "Input num_bytes exceeds 32 per element in blake3s");
         field_ct element = to_field_ct(witness_index, builder);
 
-        // byte_array_ct(field, num_bytes) constructor adds range constraints for each byte
-        byte_array_ct element_bytes(element, num_bytes);
+        // byte_array_ct(field, num_bytes) constructor adds range constraints for each byte. Note that num_bytes =
+        // ceil(witness_index_num_bits.num_bits/8). Here, num_bits is set to 8 when constructing the vector of inputs in
+        // the Blake3 constraint. Hence, we set num_bytes = 1.
+        byte_array_ct element_bytes(element, 1);
 
         // Safe write: both arr and element_bytes are constrained
         arr.write(element_bytes);
diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/gate_count_constants.hpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/gate_count_constants.hpp
index d6f71781772c..e625cf109237 100644
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/gate_count_constants.hpp
+++ b/barretenberg/cpp/src/barretenberg/dsl/acir_format/gate_count_constants.hpp
@@ -38,8 +38,8 @@ template <typename Builder> inline constexpr size_t ECDSA_SECP256K1 = 41994 + ZE
 template <typename Builder>
 inline constexpr size_t ECDSA_SECP256R1 = 72209 + ZERO_GATE + (IsMegaBuilder<Builder> ? 2 : 0);
 
-template <typename Builder> inline constexpr size_t BLAKE2S = 2959 + ZERO_GATE + MEGA_OFFSET<Builder>;
-template <typename Builder> inline constexpr size_t BLAKE3 = 2165 + ZERO_GATE + MEGA_OFFSET<Builder>;
+template <typename Builder> inline constexpr size_t BLAKE2S = 2952 + ZERO_GATE + MEGA_OFFSET<Builder>;
+template <typename Builder> inline constexpr size_t BLAKE3 = 2158 + ZERO_GATE + MEGA_OFFSET<Builder>;
 template <typename Builder> inline constexpr size_t KECCAK_PERMUTATION = 17387 + ZERO_GATE + MEGA_OFFSET<Builder>;
 template <typename Builder> inline constexpr size_t POSEIDON2_PERMUTATION = 73 + ZERO_GATE + MEGA_OFFSET<Builder>;
 template <typename Builder> inline constexpr size_t MULTI_SCALAR_MUL = 3550 + ZERO_GATE;
diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/keccak_constraint.cpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/keccak_constraint.cpp
index 41f66a3b6b6e..8d46642d2bca 100644
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/keccak_constraint.cpp
+++ b/barretenberg/cpp/src/barretenberg/dsl/acir_format/keccak_constraint.cpp
@@ -7,7 +7,6 @@
 #include "keccak_constraint.hpp"
 #include "barretenberg/stdlib/hash/keccak/keccak.hpp"
 #include "barretenberg/stdlib/primitives/circuit_builders/circuit_builders_fwd.hpp"
-#include "round.hpp"
 
 namespace acir_format {
 
diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/opcode_gate_count.test.cpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/opcode_gate_count.test.cpp
index c589795de80a..2b6e937d5980 100644
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/opcode_gate_count.test.cpp
+++ b/barretenberg/cpp/src/barretenberg/dsl/acir_format/opcode_gate_count.test.cpp
@@ -436,7 +436,7 @@ TYPED_TEST(OpcodeGateCountTests, Blake2s)
 
     blake2s_constraint.inputs.push_back(Blake2sInput{
         .blackbox_input = WitnessOrConstant<bb::fr>::from_index(0),
-        .num_bits = 32,
+        .num_bits = 8,
     });
 
     for (size_t i = 0; i < 32; ++i) {
@@ -466,7 +466,7 @@ TYPED_TEST(OpcodeGateCountTests, Blake3)
 
     blake3_constraint.inputs.push_back(Blake3Input{
         .blackbox_input = WitnessOrConstant<bb::fr>::from_index(0),
-        .num_bits = 32,
+        .num_bits = 8,
     });
 
     for (size_t i = 0; i < 32; ++i) {
diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/round.cpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/round.cpp
deleted file mode 100644
index 6e9645229981..000000000000
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/round.cpp
+++ /dev/null
@@ -1,29 +0,0 @@
-// === AUDIT STATUS ===
-// internal:    { status: Planned, auditors: [], commit: }
-// external_1:  { status: not started, auditors: [], commit: }
-// external_2:  { status: not started, auditors: [], commit: }
-// =====================
-
-#include "round.hpp"
-
-namespace acir_format {
-
-// Rounds a number to the nearest multiple of 8
-uint32_t round_to_nearest_mul_8(uint32_t num_bits)
-{
-    BB_ASSERT(num_bits <= UINT32_MAX - 8, "round_to_nearest_mul_8: num_bits too large, calculation would overflow.");
-    uint32_t remainder = num_bits % 8;
-    if (remainder == 0) {
-        return num_bits;
-    }
-
-    return num_bits + 8 - remainder;
-}
-
-// Rounds the number of bits to the nearest byte
-uint32_t round_to_nearest_byte(uint32_t num_bits)
-{
-    return round_to_nearest_mul_8(num_bits) / 8;
-}
-
-} // namespace acir_format
diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/round.hpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/round.hpp
deleted file mode 100644
index 4fc17a08211b..000000000000
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/round.hpp
+++ /dev/null
@@ -1,18 +0,0 @@
-// === AUDIT STATUS ===
-// internal:    { status: Planned, auditors: [], commit: }
-// external_1:  { status: not started, auditors: [], commit: }
-// external_2:  { status: not started, auditors: [], commit: }
-// =====================
-
-#include "barretenberg/common/assert.hpp"
-#include <cstdint>
-
-namespace acir_format {
-
-// Rounds a number to the nearest multiple of 8
-uint32_t round_to_nearest_mul_8(uint32_t num_bits);
-
-// Rounds the number of bits to the nearest byte
-uint32_t round_to_nearest_byte(uint32_t num_bits);
-
-} // namespace acir_format
diff --git a/barretenberg/cpp/src/barretenberg/dsl/acir_format/sha256_constraint.cpp b/barretenberg/cpp/src/barretenberg/dsl/acir_format/sha256_constraint.cpp
index d4ec8342235b..123b7f0d6fe2 100644
--- a/barretenberg/cpp/src/barretenberg/dsl/acir_format/sha256_constraint.cpp
+++ b/barretenberg/cpp/src/barretenberg/dsl/acir_format/sha256_constraint.cpp
@@ -7,7 +7,6 @@
 #include "sha256_constraint.hpp"
 #include "barretenberg/serialize/msgpack_impl.hpp"
 #include "barretenberg/stdlib/hash/sha256/sha256.hpp"
-#include "round.hpp"
 
 namespace acir_format {