Deploy generate_video_ltx_v1.0_blackwell but it doesn't work

[huyhuana]

I'm deploy generate_video_ltx_v1.0_blackwell to new Endpoint using Runpod console, but when test the endpoint, it's return this error.
(also error when run generate_video_client.py)

The error:
2w2wwujjoe6yfv[info]Finished.
2w2wwujjoe6yfv[error]{\n "error_type": "<class 'urllib.error.HTTPError'>",\n "error_message": "HTTP Error 400: Bad Request",\n "error_traceback": "Traceback (most recent call last):\n File "/usr/local/lib/python3.10/dist-packages/runpod/serverless/modules/rp_job.py", line 182, in run_job\n handler_return = handler(job)\n File "//handler.py", line 308, in handler\n videos = get_videos(ws, prompt)\n File "//handler.py", line 116, in get_videos\n prompt_id = queue_prompt(prompt)['prompt_id']\n File "//handler.py", line 99, in queue_prompt\n return json.loads(urllib.request.urlopen(req).read())\n File "/usr/lib/python3.10/urllib/request.py", line 216, in urlopen\n return opener.open(url, data, timeout)\n File "/usr/lib/python3.10/urllib/request.py", line 525, in open\n response = meth(req, response)\n File "/usr/lib/python3.10/urllib/request.py", line 634, in http_response\n response = self.parent.error(\n File "/usr/lib/python3.10/urllib/request.py", line 563, in error\n return self._call_chain(*args)\n File "/usr/lib/python3.10/urllib/request.py", line 496, in _call_chain\n result = func(args)\n File "/usr/lib/python3.10/urllib/request.py", line 643, in http_error_default\n raise HTTPError(req.full_url, code, msg, hdrs, fp)\nurllib.error.HTTPError: HTTP Error 400: Bad Request\n",\n "hostname": "2w2wwujjoe6yfv-644114b1",\n "worker_id": "2w2wwujjoe6yfv",\n "runpod_version": "1.8.1"\n}
2w2wwujjoe6yfv[error]Captured Handler Exception
2w2wwujjoe6yfv[info]invalid prompt: {'type': 'prompt_outputs_failed_validation', 'message': 'Prompt outputs failed validation', 'details': '', 'extra_info': {}}\n
2w2wwujjoe6yfv[info]Output will be ignored\n
2w2wwujjoe6yfv[info] - Value not in list: text_encoder: 'gemma_3_12B_it_fp8_e4m3fn.safetensors' not in ['gemma_3_12B_it_fp8_scaled.safetensors', 'ltx-2-19b-embeddings_connector_dev_bf16.safetensors']\n
2w2wwujjoe6yfv[info] LTXAVTextEncoderLoader 92:60:\n
2w2wwujjoe6yfv[info]Failed to validate prompt for output 75:\n
2w2wwujjoe6yfv[info]got prompt\n
2w2wwujjoe6yfv[info]handler.py :95 2026-02-05 06:23:26,053 Queueing prompt to: http://127.0.0.1:8188/prompt\n
2w2wwujjoe6yfv[info]handler.py :301 2026-02-05 06:23:26,053 웹소켓 연결 성공 (시도 1)\n
2w2wwujjoe6yfv[info]handler.py :286 2026-02-05 06:23:26,053 HTTP 연결 성공 (시도 1)\n
2w2wwujjoe6yfv[info]handler.py :278 2026-02-05 06:23:26,052 Checking HTTP connection to: http://127.0.0.1:8188/\n
2w2wwujjoe6yfv[info]handler.py :274 2026-02-05 06:23:26,052 Connecting to WebSocket: ws://127.0.0.1:8188/ws?clientId=cdfd17ef-655b-4097-a399-116035d978d6\n
2w2wwujjoe6yfv[info]handler.py :188 2026-02-05 06:23:26,051 Using T2V workflow: workflow/video_ltx2_t2v_renew.json\n
2w2wwujjoe6yfv[info]handler.py :186 2026-02-05 06:23:26,051 이미지 입력이 없어 T2V (Text-to-Video) 워크플로우를 사용합니다.\n
2w2wwujjoe6yfv[info]handler.py :161 2026-02-05 06:23:26,051 Received job input: {'height': 480, 'prompt': 'flying dragon', 'steps': 20, 'video_length': 49, 'width': 480}\n
2w2wwujjoe6yfv[info]Started.

[wlsdml1114]

Hi, did you connect network volume to your endpoint?

[huyhuana]

Hi, thank for reply.
I connected endpoint to network volumn (EU-RO-1), but endpoint for generate_video_ltx_v1.0_blackwell still fail.
GPU type: RTX 5090

I saw new version generate_video_ltx_v1.1_blackwell, and try to deploy new Endpoint for v1.1, but it's facing an other error:

[info]Finished.
[info]handler.py :111 2026-02-06 03:47:58,962 Getting history from: http://127.0.0.1:8188/history/9e95f741-b09c-444b-87c9-4b1910097dc1\n
[info]Prompt executed in 0.01 seconds\n
[info]\n
[info]safetensors_rust.SafetensorError: Error while deserializing header: incomplete metadata, file not fully covered\n
[info] with safetensors.safe_open(ckpt, framework="pt", device=device.type) as f:\n
[info] File "/ComfyUI/comfy/utils.py", line 120, in load_torch_file\n
[info] raise e\n
[info] File "/ComfyUI/comfy/utils.py", line 136, in load_torch_file\n
[info] sd, metadata = comfy.utils.load_torch_file(ckpt_path, return_metadata=True)\n
[info] File "/ComfyUI/comfy_extras/nodes_lt_audio.py", line 30, in execute\n
[info] to_return = cls.execute(*args, **kwargs)\n
[info] File "/ComfyUI/comfy_api/latest/_io.py", line 1710, in EXECUTE_NORMALIZED\n
[info] return method(locked_class, **inputs)\n
[info] File "/ComfyUI/comfy_api/internal/init.py", line 149, in wrapped_func\n
[info] result = f(**inputs)\n
[info] File "/ComfyUI/execution.py", line 293, in process_inputs\n
[info] await process_inputs(input_dict, i)\n
[info] File "/ComfyUI/execution.py", line 305, in _async_map_node_over_list\n
[info] return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)\n
[info] File "/ComfyUI/execution.py", line 331, in get_output_data\n
[info] output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)\n
[info] File "/ComfyUI/execution.py", line 527, in execute\n
[info]Traceback (most recent call last):\n
[info]!!! Exception during processing !!! Error while deserializing header: incomplete metadata, file not fully covered\n
[info]got prompt\n
[info]handler.py :95 2026-02-06 03:47:58,730 Queueing prompt to: http://127.0.0.1:8188/prompt\n
[info]handler.py :301 2026-02-06 03:47:58,730 웹소켓 연결 성공 (시도 1)\n
[info]handler.py :286 2026-02-06 03:47:58,729 HTTP 연결 성공 (시도 1)\n
[info]handler.py :278 2026-02-06 03:47:58,728 Checking HTTP connection to: http://127.0.0.1:8188/\n
[info]handler.py :274 2026-02-06 03:47:58,728 Connecting to WebSocket: ws://127.0.0.1:8188/ws?clientId=1c97f7aa-e207-4d93-8160-14c632a25bc9\n
[info]handler.py :188 2026-02-06 03:47:58,728 Using I2V workflow: workflow/LTX-2 I2V WORKFLOW_renew.json\n
[info]handler.py :182 2026-02-06 03:47:58,728 이미지 입력이 감지되어 I2V (Image-to-Video) 워크플로우를 사용합니다.\n
[info]handler.py :87 2026-02-06 03:47:58,728 ✅ Base64 입력을 '/task_151687a6-df3a-46be-9bbf-c0e3843ec64b/input_image.jpg' 파일로 저장했습니다.\n
[info]handler.py :45 2026-02-06 03:47:58,727 🔢 Base64 입력 처리\n
[info]mn+4yu+mKUVdh5...', 'length': 81, 'lora_pairs': [], 'negative_prompt': 'blurry, low quality, distorted', 'prompt': 'running man, grab the gun', 'seed': 42, 'steps': 10, 'width': 480}\n``