# openPangu-VL-7B-x-platform by openRFM: Open-Weight Model Source: https://savrn.com/models/openpangu-vl-7b-x-platform Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve openPangu-VL-7B-x-platform (8.8B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 17.6 GB | 21.1 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 8.8 GB | 10.5 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 4.4 GB | 5.3 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 7, 2026. [openPangu-VL-7B-x-platform on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/openpangu-vl-7b-x-platform/gpus) ## Model Card openPangu-VL-7B 是基于昇腾 NPU ，基于openPangu-Embedded-7B-V1.1语言基模和openPangu-ViT-600M视觉编码器训练的高效多模态模型。openPangu-VL-7B 训练了约 3T tokens，具备通用视觉对话、文档理解、目标定位与计数、视频理解、视觉高阶推理等能力。该模型为快思考模型。 注： 评测使用vllm-ascend部署推理，系统prompt为空。一般而言，图片最小分辨率设置为2304\28\28能获得最优的测评效果。（OCRBench中的极小图OCR除外，建议设置为不大于64\28\28。）具体prompt和分辨率设置参见技术报告附录。 - 使用vllm-ascend推理框架，参考[vllmascendforopenpanguvl7b]进行服务部署。 - python==3.10 - CANN==8.1.RC1 - 更多推理样例和能力展示，请参见cookbooks。 除文件中对开源许可证另有约定外，openPangu-VL-7B 模型根据 OPENPANGU MODEL LICENSE AGREEMENT VERSION 1.0 授权，旨在允许使用并促进人工智能技术的进一步发展。有关详细信息，请参阅模型存储库根目录中的 LICENSE 文件。 由于 openPangu-VL-7B （“模型”）所依赖的技术固有的限制，以及人工智能生成的内容是由盘古自动生成的，华为无法对以下事项做出任何保证： 如果有任何意见和建议，请提交issue或联系openPangu@huawei.com。 Excerpt from the card by openRFM. ## Configuration Architecture OpenPanguVLForConditionalGeneration Context length (tokens) 131,072 Layers 34 Hidden size 4,096 Feed-forward size 12,800 Attention heads 32 Key/value heads 8 Vocabulary size 153,376 RoPE base 6.4e+07 Stored precision bfloat16 Model type openpangu_vl ## Identity and Version Repository hk-eai/openPangu-VL-7B-x-platform Publisher openRFM Task Not stated by the source Modality Other Library Not stated by the source Parameters 8.8B parameters Languages Not stated by the source Revision 53d33a90156b8e9093e5d62676487b9cb3ab114b First published 2026-04-01 Last updated 2026-10-07 ## Files and Weights 191 files, 17.7 GB in total. The weights are 1 file totalling 17.6 GB in safetensors. Weights1 file · 17.6 GB Configuration139 files · 1.1 MB Tokenizer3 files · 2.5 MB Documentation6 files · 64.6 KB Other39 files · 176.9 MB Repository3 files · 2.2 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 17.6 GB | c3cd349fcf26 | | config.json | Configuration | 1.6 KB | — | | configuration_openpangu_vl.py | Configuration | 259 B | — | | generation_config.json | Configuration | 177 B | — | | imageprocessor_openpangu_vl.py | Configuration | 186 B | — | | inference/generate.py | Configuration | 1.8 KB | — | | inference/vllm_ascend/examples/quick_start.py | Configuration | 2.6 KB | — | | inference/vllm_ascend/pangu_infer/launcher.py | Configuration | 13.7 KB | — | | inference/vllm_ascend/pangu_infer/models/vllm_ascend/__init__.py | Configuration | 1.5 KB | — | | inference/vllm_ascend/pangu_infer/models/vllm_ascend/imageprocessor_openpangu_vl.py | Configuration | 16.1 KB | — | | inference/vllm_ascend/pangu_infer/models/vllm_ascend/modeling_openpangu_embedded.py | Configuration | 28.8 KB | — | | inference/vllm_ascend/pangu_infer/models/vllm_ascend/modeling_openpangu_vl.py | Configuration | 54.4 KB | — | | inference/vllm_ascend/pangu_infer/models/vllm_ascend/processor_openpangu_vl.py | Configuration | 10.2 KB | — | | inference/vllm_ascend/pangu_infer/models/vllm_ascend/rotary_embedding.py | Configuration | 8.2 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/kvcache/kvcache_manager.py | Configuration | 20.5 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/kvcache/swa.py | Configuration | 7.0 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/config.py | Configuration | 551 B | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/generation_monitor.py | Configuration | 33.6 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/multiproc_executor.py | Configuration | 1.2 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/repetition_detector.py | Configuration | 18.8 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/sampler.py | Configuration | 6.4 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/sequence.py | Configuration | 599 B | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/think_compressor.py | Configuration | 12.9 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/reasoning_compression/utils.py | Configuration | 1.4 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/schedule_policy/normalized_scorer.py | Configuration | 3.2 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/schedule_policy/request_queue.py | Configuration | 13.4 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/schedule_policy/schedule_policies.py | Configuration | 11.8 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/schedule_policy/weighted_score_softer.py | Configuration | 1.3 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/block_table.py | Configuration | 613 B | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/kv_cache_utils.py | Configuration | 1.7 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/no_prefix_cache_coordinator.py | Configuration | 2.6 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/sparsekv.py | Configuration | 14.3 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/sparsekv_attention_manager.py | Configuration | 1.2 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/sparsekv_block_pool_utils.py | Configuration | 1.2 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/sparsekv_runner_utils.py | Configuration | 11.3 KB | — | | inference/vllm_ascend/pangu_infer/pangu/accelerators/sparsekv/sparsekv_scheduler_utils.py | Configuration | 2.0 KB | — | | inference/vllm_ascend/pangu_infer/pangu/adaptor_manager/adaptor_manager.py | Configuration | 7.5 KB | — | | inference/vllm_ascend/pangu_infer/pangu/adaptor_manager/apply.py | Configuration | 5.1 KB | — | | inference/vllm_ascend/pangu_infer/pangu/adaptor_manager/custom_ops_patcher.py | Configuration | 7.5 KB | — | | inference/vllm_ascend/pangu_infer/pangu/adaptor_manager/router_patcher.py | Configuration | 4.1 KB | — | | inference/vllm_ascend/pangu_infer/pangu/utils/pangu_config.py | Configuration | 1.3 KB | — | | inference/vllm_ascend/pangu_infer/pangu/utils/utils.py | Configuration | 1.3 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/__init__.py | Configuration | 427 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm/attention/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/attention/layer.py | Configuration | 5.5 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/config.py | Configuration | 15.2 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/engine/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/engine/protocol.py | Configuration | 538 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/chat_utils.py | Configuration | 4.8 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/api_server.py | Configuration | 4.3 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/protocol.py | Configuration | 12.5 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/reasoning_parsers/__init__.py | Configuration | 93 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/reasoning_parsers/pangu_reasoning_parser.py | Configuration | 7.2 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/serving_chat.py | Configuration | 8.9 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/serving_completion.py | Configuration | 11.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/tool_parsers/__init__.py | Configuration | 78 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm/entrypoints/openai/tool_parsers/pangu_tool_parser.py | Configuration | 13.4 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/model_executor/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/model_executor/layers/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/model_executor/layers/fused_moe/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/model_executor/layers/fused_moe/layer.py | Configuration | 13.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/model_executor/layers/linear.py | Configuration | 1.9 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/model_executor/layers/rotary_embedding.py | Configuration | 17.6 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/multimodal/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/multimodal/utils.py | Configuration | 1.1 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/multimodal/video.py | Configuration | 5.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/sampling_params.py | Configuration | 641 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm/transformers_utils/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/core/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/core/block_pool.py | Configuration | 1.7 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/core/kv_cache_coordinator.py | Configuration | 2.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/core/kv_cache_utils.py | Configuration | 3.1 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/core/sched/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/core/sched/scheduler.py | Configuration | 38.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/core/single_type_kv_cache_manager.py | Configuration | 1.2 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/engine/__init__.py | Configuration | 1.8 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/engine/core.py | Configuration | 1.2 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/engine/core_client.py | Configuration | 5.1 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/engine/processor.py | Configuration | 6.5 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/executor/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/executor/abstract.py | Configuration | 2.3 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/request.py | Configuration | 5.2 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/sample/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/sample/metadata.py | Configuration | 987 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/sample/ops/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/sample/ops/topk_topp_sampler.py | Configuration | 2.1 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/sample/rejection_sampler.py | Configuration | 3.4 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/v1/sample/sampler.py | Configuration | 3.4 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/worker/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm/worker/gpu_input_batch.py | Configuration | 33.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm/worker/worker_base.py | Configuration | 4.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/__init__.py | Configuration | 427 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/ascend_config.py | Configuration | 6.7 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/attention/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/attention/attention_v1.py | Configuration | 34.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/attention/attention_v1_torchair.py | Configuration | 35.5 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/attention/mla_v1.py | Configuration | 6.3 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/distributed/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/distributed/llmdatadist_c_mgr_connector.py | Configuration | 4.1 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/distributed/parallel_state.py | Configuration | 3.7 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/ops/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/ops/attention.py | Configuration | 6.6 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/ops/common_fused_moe.py | Configuration | 3.4 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/ops/fused_moe.py | Configuration | 25.6 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/ops/rotary_embedding.py | Configuration | 1.3 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/platform.py | Configuration | 10.5 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/quantization/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/quantization/quant_config.py | Configuration | 3.6 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/quantization/w8a8.py | Configuration | 1.8 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/quantization/w8a8_dynamic.py | Configuration | 23.6 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/sample/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/sample/rejection_sampler.py | Configuration | 9.5 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/worker/__init__.py | Configuration | | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/worker/cache_engine.py | Configuration | 3.5 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/worker/model_runner.py | Configuration | 3.6 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/worker/model_runner_v1.py | Configuration | 92.0 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/worker/mtp_proposer_v1.py | Configuration | 18.8 KB | — | | inference/vllm_ascend/pangu_infer/patches/vllm_ascend/worker/worker_v1.py | Configuration | 6.1 KB | — | | inference/vllm_ascend/pangu_infer/sitecustomize.py | Configuration | 2.8 KB | — | | main.py | Configuration | 8.6 KB | — | | model.safetensors.index.json | Configuration | 46.4 KB | — | | modeling_openpangu_embedded.py | Configuration | 261 B | — | | modeling_openpangu_vl.py | Configuration | 249 B | — | | openpangu_vl/__init__.py | Configuration | 355 B | — | | openpangu_vl/configuration_openpangu_vl.py | Configuration | 5.7 KB | — | | openpangu_vl/imageprocessor_openpangu_vl.py | Configuration | 16.8 KB | — | | openpangu_vl/modeling_openpangu_embedded.py | Configuration | 31.1 KB | — | | openpangu_vl/modeling_openpangu_vl.py | Configuration | 88.7 KB | — | | openpangu_vl/processor_openpangu_vl.py | Configuration | 10.2 KB | — | | openpangu_vl/tokenization_openpangu.py | Configuration | 10.5 KB | — | | openpangu_vl/videoprocessor_openpangu_vl.py | Configuration | 8.1 KB | — | | preprocessor_config.json | Configuration | 823 B | — | | processor_openpangu_vl.py | Configuration | 251 B | — | | special_tokens_map.json | Configuration | 414 B | — | | tokenization_openpangu.py | Configuration | 181 B | — | | video_preprocessor_config.json | Configuration | 1.4 KB | — | | videoprocessor_openpangu_vl.py | Configuration | 261 B | — | | LICENSE | Documentation | 4.5 KB | — | | OPEN SOURCE SOFTWARE NOTICE | Documentation | 35.5 KB | — | | README.md | Documentation | 5.0 KB | — | | README_EN.md | Documentation | 5.7 KB | — | | doc/vllm_ascend_for_openpangu_vl_7b.md | Documentation | 6.7 KB | — | | doc/vllm_ascend_for_openpangu_vl_7b_EN.md | Documentation | 7.2 KB | — | | chat_template.jinja | Other | 1.4 KB | — | | cookbooks/assets/grounding/depth_1.jpg | Other | 109.2 KB | 21ac7fa907d3 | | cookbooks/assets/grounding/depth_2.jpg | Other | 131.4 KB | ea6caad976a5 | | cookbooks/assets/grounding/dinner.jpg | Other | 440.9 KB | bd70e19ff10d | | cookbooks/assets/grounding/macaron.jpg | Other | 116.7 KB | 9604a04be51e | | cookbooks/assets/grounding/peoples.jpg | Other | 101.1 KB | 3d49bf9200e7 | | cookbooks/assets/grounding/test_example_point_01.png | Other | 1.1 MB | 4a65d21bb6d0 | | cookbooks/assets/grounding/test_example_point_02.png | Other | 545.5 KB | 7cbeff1c16bf | | cookbooks/assets/grounding/tools.jpg | Other | 69.5 KB | d9843a273c47 | | cookbooks/assets/ocr/example1_1.png | Other | 536.3 KB | a00ab06d34e0 | | cookbooks/assets/ocr/example1_2.png | Other | 317.5 KB | 8a1e06f03f8c | | cookbooks/assets/ocr/example2.png | Other | 594.1 KB | 7d2e6ba81924 | | cookbooks/assets/ocr/example3.jpg | Other | 74.8 KB | — | | cookbooks/assets/ocr/example4.png | Other | 111.3 KB | 1b2c35cfff31 | | cookbooks/assets/ocr/example5.png | Other | 1.3 MB | e8e6388d05d1 | | cookbooks/assets/ocr/example6.png | Other | 121.4 KB | 079d2df93107 | | cookbooks/assets/ocr/example7.png | Other | 292.5 KB | b514ab9dd987 | | cookbooks/assets/ocr/example7_2.png | Other | 233.4 KB | 01ae75907ed4 | | cookbooks/assets/reasoning/biology.png | Other | 7.8 KB | 03d5588406cc | | cookbooks/assets/reasoning/chemistry.png | Other | 152.1 KB | 73c330872e6c | | cookbooks/assets/reasoning/geometry.png | Other | 264.9 KB | d91e111372ff | | cookbooks/assets/reasoning/logical.png | Other | 8.3 KB | 009b3d7a457c | | cookbooks/assets/video/example_video_1.mp4 | Other | 12.7 MB | 7a39adc894b3 | | cookbooks/assets/video/example_video_2.mp4 | Other | 14.9 MB | 2a90a370b6a7 | | cookbooks/assets/video/example_video_3.mp4 | Other | 26.4 MB | a005fdc45c1e | | cookbooks/assets/video/example_video_4.mp4 | Other | 8.1 MB | cc2a87bf37fe | | cookbooks/assets/video/example_video_5.mp4 | Other | 4.8 MB | 0613a6189fb5 | | cookbooks/assets/video/example_video_6.mp4 | Other | 19.9 MB | 2bd56ba8ef4d | | cookbooks/assets/video/example_video_7.mp4 | Other | 23.5 MB | f3020815b636 | | cookbooks/assets/video/example_video_8.mp4 | Other | 25.9 MB | cf41e2c4a511 | | cookbooks/assets/video/example_video_9.mp4 | Other | 16.1 MB | 121f77f79e28 | | cookbooks/grounding.ipynb | Other | 10.7 MB | 68a1d3e9b017 | | cookbooks/ocr.ipynb | Other | 5.5 MB | — | | cookbooks/reasoning.ipynb | Other | 605.2 KB | — | | cookbooks/video.ipynb | Other | 12.2 KB | — | | doc/technical_report.pdf | Other | 1.3 MB | 3d9b7f7fd871 | | inference/requirements.txt | Other | 71 B | — | | inference/vllm_ascend/examples/start_serving_openpangu_vl_7b.sh | Other | 3.5 KB | — | | pyproject.toml | Other | 736 B | — | | .gitattributes | Repository | 2.1 KB | — | | .gitignore | Repository | 129 B | — | | .python-version | Repository | 5 B | — | | inference/vllm_ascend/pangu_infer/patches/vllm/transformers_utils/tokenizer_group.py | Tokenizer | 1.3 KB | — | | tokenizer.model | Tokenizer | 2.5 MB | 6b16f1558c0c | | tokenizer_config.json | Tokenizer | 10.6 KB | — | ## License and Download License Not stated by the source Access Open weights, no gate Download size 17.6 GB [Download from openRFM](https://huggingface.co/hk-eai/openPangu-VL-7B-x-platform) Released by openRFM through its official repository on Hugging Face. ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 17.6 GB | | 16-bit | 17.6 GB | | 8-bit | 8.8 GB | | 4-bit | 4.4 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About openPangu-VL-7B-x-platform ### How much GPU memory does openPangu-VL-7B-x-platform need? About 21.1 GB at 16-bit and 5.3 GB at 4-bit: the weights (8.8B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run openPangu-VL-7B-x-platform on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### What is openPangu-VL-7B-x-platform's context length? 131,072 tokens, from the maximum position embeddings in its published configuration. ## openRFM [All models and datasets](https://savrn.com/model-publishers/hk-eai) ## Versions - [53d33a90156b](https://savrn.com/models/openpangu-vl-7b-x-platform/versions/53d33a90156b) · current 2026-10-07 ## Explore More - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-07. - [Hugging Face record](https://huggingface.co/hk-eai/openPangu-VL-7B-x-platform) - [How the hub is built](https://savrn.com/model-hub/methodology)