diff --git a/ggml/include/ggml.h b/ggml/include/ggml.h
index 9c4e24023b5ad..cfd88caf34367 100644
--- a/ggml/include/ggml.h
+++ b/ggml/include/ggml.h
@@ -481,6 +481,7 @@ extern "C" {
         GGML_OP_CONV_TRANSPOSE_1D,
         GGML_OP_IM2COL,
         GGML_OP_IM2COL_BACK,
+        GGML_OP_CONV_2D,
         GGML_OP_CONV_2D_DW,
         GGML_OP_CONV_TRANSPOSE_2D,
         GGML_OP_POOL_1D,
@@ -1723,6 +1724,17 @@ extern "C" {
             struct ggml_tensor  * b,
             int                   stride);
 
+    GGML_API struct ggml_tensor * ggml_conv_2d_direct(
+            struct ggml_context * ctx,
+            struct ggml_tensor  * a,   // convolution kernel [KW, KH, IC, OC]
+            struct ggml_tensor  * b,   // input data [W, H, C, N]
+            int                   s0,  // stride dimension 0
+            int                   s1,  // stride dimension 1
+            int                   p0,  // padding dimension 0
+            int                   p1,  // padding dimension 1
+            int                   d0,  // dilation dimension 0
+            int                   d1); // dilation dimension 1
+
     enum ggml_op_pool {
         GGML_OP_POOL_MAX,
         GGML_OP_POOL_AVG,
diff --git a/ggml/src/ggml-cpu/ggml-cpu.c b/ggml/src/ggml-cpu/ggml-cpu.c
index 1d3cd009affc6..14a1934e06d84 100644
--- a/ggml/src/ggml-cpu/ggml-cpu.c
+++ b/ggml/src/ggml-cpu/ggml-cpu.c
@@ -683,6 +683,10 @@ static void ggml_init_arm_arch_features(void) {
 
 #endif // __ARM_ARCH
 
+void ggml_compute_forward_mul_mat(
+        const struct ggml_compute_params * params,
+              struct ggml_tensor * dst);
+
 struct ggml_tensor * ggml_new_i32(struct ggml_context * ctx, int32_t value) {
     GGML_ASSERT(!ggml_get_no_alloc(ctx));
 
@@ -1189,7 +1193,7 @@ static void ggml_compute_forward_mul_mat_one_chunk(
     }
 }
 
-static void ggml_compute_forward_mul_mat(
+void ggml_compute_forward_mul_mat(
         const struct ggml_compute_params * params,
               struct ggml_tensor * dst) {
 
@@ -1858,6 +1862,10 @@ static void ggml_compute_forward(struct ggml_compute_params * params, struct ggm
             {
                 ggml_compute_forward_im2col_back_f32(params, tensor);
             } break;
+        case GGML_OP_CONV_2D:
+            {
+                ggml_compute_forward_conv_2d(params, tensor);
+            } break;
         case GGML_OP_CONV_2D_DW:
             {
                 ggml_compute_forward_conv_2d_dw(params, tensor);
@@ -2203,6 +2211,7 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
             } break;
         case GGML_OP_IM2COL:
         case GGML_OP_IM2COL_BACK:
+        case GGML_OP_CONV_2D:
         case GGML_OP_CONV_2D_DW:
         case GGML_OP_CONV_TRANSPOSE_1D:
         case GGML_OP_CONV_TRANSPOSE_2D:
@@ -2721,6 +2730,12 @@ struct ggml_cplan ggml_graph_plan(
                             GGML_ABORT("fatal error");
                         }
                     } break;
+                case GGML_OP_CONV_2D:
+                    {
+                        cur = GGML_IM2COL_WORK_SIZE;
+                        //Add enough space for kernel transpose
+                        cur += sizeof(ggml_fp16_t)*node->src[1]->ne[0]*node->src[1]->ne[1]*node->src[1]->ne[2]*node->src[1]->ne[3];
+                    } break;
                 case GGML_OP_CONV_TRANSPOSE_2D:
                     {
                         const int64_t ne00 = node->src[0]->ne[0]; // W
diff --git a/ggml/src/ggml-cpu/ops.cpp b/ggml/src/ggml-cpu/ops.cpp
index eff4a53e3442b..5f55f5f8f8788 100644
--- a/ggml/src/ggml-cpu/ops.cpp
+++ b/ggml/src/ggml-cpu/ops.cpp
@@ -3,6 +3,7 @@
 #include "ggml-cpu.h"
 #include "ggml-impl.h"
 #include "binary-ops.h"
+#include "ggml.h"
 #include "unary-ops.h"
 #include "vec.h"
 
@@ -6058,6 +6059,180 @@ void ggml_compute_forward_im2col_back_f32(
     }
 }
 
+static void ggml_call_mul_mat(
+    const ggml_compute_params * params,
+    int64_t m, int64_t n, int64_t k,
+    void * a, void * b, void * c) {
+
+    struct ggml_tensor src1 = {};
+    src1.ne[0] = k;
+    src1.ne[1] = m;
+    src1.ne[2] = 1;
+    src1.ne[3] = 1;
+    src1.nb[0] = sizeof(float);
+    src1.nb[1] = k * sizeof(float);
+    src1.nb[2] = src1.nb[1];
+    src1.nb[3] = src1.nb[2];
+    src1.data  = a;
+
+    struct ggml_tensor src0 = {};
+    src0.ne[0] = k;
+    src0.ne[1] = n;
+    src0.ne[2] = 1;
+    src0.ne[3] = 1;
+    src0.nb[0] = sizeof(float);
+    src0.nb[1] = k * sizeof(float);
+    src0.nb[2] = src0.nb[1];
+    src0.nb[3] = src0.nb[2];
+    src0.data  = b;
+
+    struct ggml_tensor dst = {};
+    dst.ne[0] = n;
+    dst.ne[1] = m;
+    dst.ne[2] = 1;
+    dst.ne[3] = 1;
+    dst.nb[0] = sizeof(float);
+    dst.nb[1] = n * sizeof(float);
+    dst.nb[2] = dst.nb[1];
+    dst.nb[3] = dst.nb[2];
+    dst.data  = c;
+    dst.src[0] = &src0;
+    dst.src[1] = &src1;
+
+    ggml_compute_forward_mul_mat(params, &dst);
+}
+
+
+// ggml_compute_forward_conv_2d
+
+static void ggml_compute_forward_conv_2d_f32(
+        const ggml_compute_params * params,
+        const ggml_tensor * kernel,  // [KW, KH, IC, OC] - fp32
+        const ggml_tensor * src,     // [W, H, C, N]
+        ggml_tensor * dst) {         // [OW, OH, OC, N]
+
+    GGML_ASSERT(ggml_is_contiguous(kernel));
+    GGML_ASSERT(kernel->type == GGML_TYPE_F32);
+
+    const int32_t stride_x   = dst->op_params[0];
+    const int32_t stride_y   = dst->op_params[1];
+    const int32_t pad_x      = dst->op_params[2];
+    const int32_t pad_y      = dst->op_params[3];
+    const int32_t dilation_x = dst->op_params[4];
+    const int32_t dilation_y = dst->op_params[5];
+
+    const int64_t c_in  = src->ne[2];
+    const int64_t c_out = kernel->ne[3];
+    GGML_ASSERT(c_in == kernel->ne[2]);
+
+    const int64_t src_w = src->ne[0];
+    const int64_t src_h = src->ne[1];
+    const int64_t knl_w = kernel->ne[0];
+    const int64_t knl_h = kernel->ne[1];
+    const int64_t dst_w = dst->ne[0];
+    const int64_t dst_h = dst->ne[1];
+
+    float * src_data = (float*) src->data;
+    float * knl_data = (float*) kernel->data;
+    float * dst_data = (float*) dst->data;
+
+    const int64_t knl_n           = knl_w * knl_h * c_in;
+    const int64_t patch_total     = dst->ne[3] * dst_w * dst_h;
+
+    const int64_t space_per_patch   = knl_n * sizeof(float) + c_out * sizeof(float);
+    const int64_t batch_size        = params->wsize / space_per_patch;
+    const int64_t patches_per_batch = batch_size > 8 ? (batch_size / 8) * 8 : batch_size;
+    const int64_t batch_n           = (patch_total + patches_per_batch - 1) / patches_per_batch;
+
+    GGML_ASSERT(patches_per_batch > 0 && batch_size >= 1);
+
+    float * tmp = (float *) params->wdata;
+
+    for (int64_t batch_i = 0; batch_i < batch_n; ++batch_i) {
+
+        const int64_t patch_start_batch = batch_i * patches_per_batch;
+        const int64_t patch_end_batch   = std::min(patch_start_batch + patches_per_batch,
+                                              patch_total);
+        const int64_t patch_n           = patch_end_batch - patch_start_batch;
+
+        const int64_t patch_per_thread  = (patch_n + params->nth - 1) / params->nth;
+        const int64_t patch_start       = patch_start_batch + params->ith * patch_per_thread;
+        const int64_t patch_end         = std::min(patch_start + patch_per_thread,patch_end_batch);
+
+        //im2col for a patch
+        for (int64_t p = patch_start; p < patch_end; ++p) {
+            const int64_t  batch_n     =  p / (dst_w * dst_h);
+            const int64_t  src_x       = (p / dst_w) % dst_h;
+            const int64_t  src_y       =  p % dst_w;
+
+            float * src_base = (float *)((char *)src_data + batch_n * src->nb[3]);
+            float * dst_row  = tmp + (p % patches_per_batch) * knl_n;
+
+            for (int64_t ic = 0; ic < c_in; ++ic) {
+                for (int64_t ky = 0; ky < knl_h; ++ky) {
+                    for (int64_t kx = 0; kx < knl_w; ++kx) {
+                        const int64_t sy = src_x * stride_y + ky * dilation_y - pad_y;
+                        const int64_t sx = src_y * stride_x + kx * dilation_x - pad_x;
+
+                        int64_t dst_idx = ic * (knl_h * knl_w) + ky * knl_w + kx;
+
+                        if (sy < 0 || sy >= src_h || sx < 0 || sx >= src_w) {
+                            dst_row[dst_idx] = 0.0f;
+                        } else {
+                            float * src_ptr = (float *)((char *)src_base + sx * src->nb[0] + sy * src->nb[1] + ic * src->nb[2]);
+                            dst_row[dst_idx] = *src_ptr;
+                        }
+                    }
+                }
+            }
+        }   // patches handled by this thread
+
+        ggml_barrier(params->threadpool);
+
+        float * gemm_output = tmp + patches_per_batch * knl_n;
+
+        // GEMM: patches[patch_n, knl_n] × kernel[knl_n, c_out] = output[patch_n, c_out]
+        ggml_call_mul_mat(params, patch_n, c_out, knl_n,
+                            tmp, knl_data, gemm_output);
+
+        ggml_barrier(params->threadpool);
+
+
+        //permute back [OC, N, OH, OW] to [N, OC, OH, OW]
+        const int64_t permute_per_thread = (patch_n + params->nth - 1) / params->nth;
+        const int64_t permute_start = params->ith * permute_per_thread;
+        const int64_t permute_end = std::min(permute_start + permute_per_thread, patch_n);
+
+        for (int64_t i = permute_start; i < permute_end; ++i) {
+            const int64_t p       = patch_start_batch + i;
+            const int64_t batch_n = p / (dst_w * dst_h);
+            const int64_t dst_y   = (p / dst_w) % dst_h;
+            const int64_t dst_x   = p % dst_w;
+
+            for (int64_t oc = 0; oc < c_out; ++oc) {
+                const float value = gemm_output[i * c_out + oc];
+                // Write to WHCN layout: dst[w, h, c, n]
+                float * dst_ptr = (float *)((char *)dst_data + dst_x * dst->nb[0] + dst_y * dst->nb[1] + oc * dst->nb[2] + batch_n * dst->nb[3]);
+                *dst_ptr = value;
+            }
+        }
+    }
+}
+
+void ggml_compute_forward_conv_2d(
+        const ggml_compute_params * params,
+        ggml_tensor * dst) {
+
+    const ggml_tensor * src0 = dst->src[0];
+    const ggml_tensor * src1 = dst->src[1];
+
+    if (src0->type == GGML_TYPE_F16) {
+        GGML_ASSERT(false && "F16 not supported yet");
+    } else {
+        ggml_compute_forward_conv_2d_f32(params, src0, src1, dst);
+    }
+}
+
 // ggml_compute_forward_conv_transpose_2d
 
 void ggml_compute_forward_conv_transpose_2d(
diff --git a/ggml/src/ggml-cpu/ops.h b/ggml/src/ggml-cpu/ops.h
index 2d8544d7d3d43..6bcbb0a2fab88 100644
--- a/ggml/src/ggml-cpu/ops.h
+++ b/ggml/src/ggml-cpu/ops.h
@@ -20,6 +20,9 @@
 
 static const size_t CACHE_LINE_SIZE_F32 = CACHE_LINE_SIZE/sizeof(float);
 
+// Work buffer size for im2col operations in CONV2D
+#define GGML_IM2COL_WORK_SIZE (16 * 1024 * 1024)  // 16MB work buffer
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -64,6 +67,7 @@ void ggml_compute_forward_clamp(const struct ggml_compute_params * params, struc
 void ggml_compute_forward_conv_transpose_1d(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_im2col(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_im2col_back_f32(const struct ggml_compute_params * params, struct ggml_tensor * dst);
+void ggml_compute_forward_conv_2d(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_conv_transpose_2d(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_conv_2d_dw(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_pool_1d(const struct ggml_compute_params * params, struct ggml_tensor * dst);
@@ -105,6 +109,7 @@ void ggml_compute_forward_custom(const struct ggml_compute_params * params, stru
 void ggml_compute_forward_cross_entropy_loss(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_cross_entropy_loss_back(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_opt_step_adamw(const struct ggml_compute_params * params, struct ggml_tensor * dst);
+void ggml_compute_forward_mul_mat(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 
 #ifdef __cplusplus
 }
diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c
index f8e7c595bce15..db562bdf53f96 100644
--- a/ggml/src/ggml.c
+++ b/ggml/src/ggml.c
@@ -986,7 +986,7 @@ static const char * GGML_OP_NAME[GGML_OP_COUNT] = {
     "OPT_STEP_ADAMW",
 };
 
-static_assert(GGML_OP_COUNT == 83, "GGML_OP_COUNT != 83");
+static_assert(GGML_OP_COUNT == 84, "GGML_OP_COUNT != 84");
 
 static const char * GGML_OP_SYMBOL[GGML_OP_COUNT] = {
     "none",
@@ -1043,6 +1043,7 @@ static const char * GGML_OP_SYMBOL[GGML_OP_COUNT] = {
     "conv_transpose_1d(x)",
     "im2col(x)",
     "im2col_back(x)",
+    "conv_2d(x)",
     "conv_2d_dw(x)",
     "conv_transpose_2d(x)",
     "pool_1d(x)",
@@ -1082,7 +1083,7 @@ static const char * GGML_OP_SYMBOL[GGML_OP_COUNT] = {
     "adamw(x)",
 };
 
-static_assert(GGML_OP_COUNT == 83, "GGML_OP_COUNT != 83");
+static_assert(GGML_OP_COUNT == 84, "GGML_OP_COUNT != 84");
 
 static_assert(GGML_OP_POOL_COUNT == 2, "GGML_OP_POOL_COUNT != 2");
 
@@ -4131,6 +4132,44 @@ struct ggml_tensor * ggml_conv_2d_dw_direct(
     return result;
 }
 
+// ggml_conv_2d_direct
+
+struct ggml_tensor * ggml_conv_2d_direct(
+        struct ggml_context * ctx,
+        struct ggml_tensor  * a,   // convolution kernel [KW, KH, IC, OC]
+        struct ggml_tensor  * b,   // input data [W, H, C, N]
+        int                   s0,  // stride dimension 0
+        int                   s1,  // stride dimension 1
+        int                   p0,  // padding dimension 0
+        int                   p1,  // padding dimension 1
+        int                   d0,  // dilation dimension 0
+        int                   d1) {// dilation dimension 1
+
+    GGML_ASSERT(a->ne[2] == b->ne[2]);
+    //GGML_ASSERT(a->type == b->type);
+
+    int64_t ne[4];
+    ne[0] = ggml_calc_conv_output_size(b->ne[0], a->ne[0], s0, p0, d0);
+    ne[1] = ggml_calc_conv_output_size(b->ne[1], a->ne[1], s1, p1, d1);
+    ne[2] = a->ne[3];
+    ne[3] = b->ne[3];
+
+    struct ggml_tensor * result = ggml_new_tensor(ctx, b->type, 4, ne);
+
+    ggml_set_op_params_i32(result, 0, s0);
+    ggml_set_op_params_i32(result, 1, s1);
+    ggml_set_op_params_i32(result, 2, p0);
+    ggml_set_op_params_i32(result, 3, p1);
+    ggml_set_op_params_i32(result, 4, d0);
+    ggml_set_op_params_i32(result, 5, d1);
+
+    result->op = GGML_OP_CONV_2D;
+    result->src[0] = a;
+    result->src[1] = b;
+
+    return result;
+}
+
 // ggml_conv_transpose_2d_p0
 
 static int64_t ggml_calc_conv_transpose_output_size(int64_t ins, int64_t ks, int s, int p) {