GCVE-110-OSM-2026-3028

Malicious package detected. Behaviors: data exfiltration, code execution, network activity, obfuscated code. Payload: xinference/model/llm/sglang/core.py Secondary files: xinference/thirdparty/deepseek_vl2/serve/app_modules/utils.py, xinference/core/worker.py Key findings: - Corporate Environment Targeting in xinference/model/llm/sglang/core.py: "tModel(SGLANGModel, ChatModelMixin): @classmethod def match" - Corporate Environment Targeting in xinference/model/llm/transformers/gemma3.py: "tModel(PytorchChatModel): GEMMA3_ARCHITECTURES = {"Gemma3ForCausalLM"} ..." - Corporate Environment Targeting in xinference/model/llm/vllm/core.py: "tModel(VLLMModel, ChatModelMixin): @classmethod def match" - Corporate Environment Targeting in xinference/thirdparty/fish_speech/fish_speech/utils/braceexpand.py: "tmod(optionflags=doctest" - Corporate Environment Targeting in xinference/thirdparty/indextts/gpt/transformers_modeling_utils.py: "tModel.from_pretrained("./test" IOCs: - ipv4: 123.0.0.0 - ipv6: 0::, 1::, 2::, 4:: - urls: http://www.apache.org/licenses/, http://www.apache.org/licenses/LICENSE-2.0, https://xinference.io/en, https://hub.docker.com/r/xprobe/xinference, https://bfl.ai/models/flux-2-klein (+45 more) - domains: www.apache.org, xinference.io, actions-badge.atrox.dev, hub.docker.com, huggingface.co (+27 more) - emails: qinxuye@xprobe.io, business@xprobe.io, developers@01.AI, sxc19@mails.tsinghua.edu.cn, zyzhou@stu.xmu.edu.cn (+1 more) - bitcoinAddresses: 18wF2bbjgc3ZzCSKW1T5nt5EbFoAz, 31WkbgnQpgtrNp1SdpJvpUAGTa - sha256Hashes: d3dd57d32accea0b295c96e26691aa14d8822fac7d9d27d5dc00b4ca2826dd03, 65147644a518d12f04e32d6f3b26facc3f8dd46e5390956a9424a650c0ce22b9, 25a8566e1d0c1e2231d1c762132cd20e0f96a85d16145c3a00adf5d1ac670ead, ed3a0b6b1c0edf879ad9b11b1af5a0e6ab5db9205f891f668f8b0e6c6326e34e, f953ad0fd29cacd07d5a9eda5624af0f6bcf2258be67c92b79389873d91e0872 (+6 more) - payloadFileHash: 52452de76b8e1da247af3cb2673be345844828ecf8a4bc3be07c9adc0f3fd417