diff --git a/README.md b/README.md
index f3ab7782..073a841d 100644
--- a/README.md
+++ b/README.md
@@ -247,7 +247,6 @@ Optionally, you can use the following command-line flags:
 | `--mlock`   | Force the system to keep the model in RAM. |
 | `--cache-capacity CACHE_CAPACITY`   | Maximum cache capacity. Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed. |
 | `--n-gpu-layers N_GPU_LAYERS` | Number of layers to offload to the GPU. Only works if llama-cpp-python was compiled with BLAS. Set this to 1000000000 to offload all layers to the GPU. |
-| `--tensor_split TENSOR_SPLIT` | Split the model across multiple GPUs, comma-separated list of proportions, e.g. 18,17 |
 | `--n_ctx N_CTX` | Size of the prompt context. |
 | `--llama_cpp_seed SEED` | Seed for llama-cpp models. Default 0 (random). |
 | `--n_gqa N_GQA`         | grouped-query attention. Must be 8 for llama2 70b. |
diff --git a/modules/llamacpp_hf.py b/modules/llamacpp_hf.py
index 5c8c1a7a..6a4a0294 100644
--- a/modules/llamacpp_hf.py
+++ b/modules/llamacpp_hf.py
@@ -94,12 +94,6 @@ class LlamacppHF(PreTrainedModel):
             model_file = list(path.glob('*ggml*.bin'))[0]
 
         logger.info(f"llama.cpp weights detected: {model_file}\n")
-
-        if shared.args.tensor_split is None or shared.args.tensor_split.strip() == '':
-            tensor_split_list = None
-        else:
-            tensor_split_list = [float(x) for x in shared.args.tensor_split.strip().split(",")]
-
         params = {
             'model_path': str(model_file),
             'n_ctx': shared.args.n_ctx,
@@ -110,7 +104,6 @@ class LlamacppHF(PreTrainedModel):
             'use_mlock': shared.args.mlock,
             'low_vram': shared.args.low_vram,
             'n_gpu_layers': shared.args.n_gpu_layers,
-            'tensor_split': tensor_split_list,
             'rope_freq_base': 10000 * shared.args.alpha_value ** (64/63.),
             'rope_freq_scale': 1.0 / shared.args.compress_pos_emb,
             'n_gqa': shared.args.n_gqa or None,
diff --git a/modules/llamacpp_model.py b/modules/llamacpp_model.py
index 69e1c160..0f9c3470 100644
--- a/modules/llamacpp_model.py
+++ b/modules/llamacpp_model.py
@@ -41,12 +41,6 @@ class LlamaCppModel:
                 cache_capacity = int(shared.args.cache_capacity)
 
         logger.info("Cache capacity is " + str(cache_capacity) + " bytes")
-
-        if shared.args.tensor_split is None or shared.args.tensor_split.strip() == '':
-            tensor_split_list = None
-        else:
-            tensor_split_list = [float(x) for x in shared.args.tensor_split.strip().split(",")]
-
         params = {
             'model_path': str(path),
             'n_ctx': shared.args.n_ctx,
@@ -57,7 +51,6 @@ class LlamaCppModel:
             'use_mlock': shared.args.mlock,
             'low_vram': shared.args.low_vram,
             'n_gpu_layers': shared.args.n_gpu_layers,
-            'tensor_split': tensor_split_list,
             'rope_freq_base': 10000 * shared.args.alpha_value ** (64/63.),
             'rope_freq_scale': 1.0 / shared.args.compress_pos_emb,
             'n_gqa': shared.args.n_gqa or None,
diff --git a/modules/loaders.py b/modules/loaders.py
index acb59c65..c55cf0ff 100644
--- a/modules/loaders.py
+++ b/modules/loaders.py
@@ -33,7 +33,6 @@ loaders_and_params = {
         'n_gqa',
         'rms_norm_eps',
         'n_gpu_layers',
-        'tensor_split',
         'n_batch',
         'threads',
         'no_mmap',
@@ -48,7 +47,6 @@ loaders_and_params = {
         'n_gqa',
         'rms_norm_eps',
         'n_gpu_layers',
-        'tensor_split',
         'n_batch',
         'threads',
         'no_mmap',
diff --git a/modules/shared.py b/modules/shared.py
index 978a818c..a0402878 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -125,7 +125,6 @@ parser.add_argument('--low-vram', action='store_true', help='Low VRAM Mode')
 parser.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.')
 parser.add_argument('--cache-capacity', type=str, help='Maximum cache capacity. Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed.')
 parser.add_argument('--n-gpu-layers', type=int, default=0, help='Number of layers to offload to the GPU.')
-parser.add_argument('--tensor_split', type=str, default=None, help="Split the model across multiple GPUs, comma-separated list of proportions, e.g. 18,17")
 parser.add_argument('--n_ctx', type=int, default=2048, help='Size of the prompt context.')
 parser.add_argument('--llama_cpp_seed', type=int, default=0, help='Seed for llama-cpp models. Default 0 (random)')
 parser.add_argument('--n_gqa', type=int, default=0, help='grouped-query attention. Must be 8 for llama2 70b.')
diff --git a/modules/ui.py b/modules/ui.py
index 3e466ffa..d9b3a131 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -60,7 +60,6 @@ def list_model_elements():
         'low_vram',
         'mlock',
         'n_gpu_layers',
-        'tensor_split',
         'n_ctx',
         'n_gqa',
         'rms_norm_eps',