Security Advisory

CVE-2026-73557

CVE vulnerability detail - eXtreme Datacenter Security Operations

Published 2026-08-13 15:00:15

Last updated 2026-08-13 17:52:52

Assigner GitHub_M

CVSS score not scored

State PUBLISHED

Description

vLLM is an inference and serving engine for large language models. From 0.20.2rc0 until 0.26.0, safe_load_prompt_embeds in vllm/renderers/embed_utils.py uses torch.sparse.check_sparse_tensor_invariants, whose process-global save, enable, and restore state can be raced by concurrent prompt_embeds parts submitted to POST /v1/chat/completions through AsyncMultiModalItemTracker.resolve_items, asyncio.gather, and the default executor, allowing an invalid sparse tensor to reach tensor.to_dense despite the CVE-2025-62164 guard when enable_prompt_embeds is enabled. This issue is fixed in version 0.26.0.

← Browse all CVEs View on cve.org ↗