CVE-2026-73557
Overview
CVE-2026-73557 is a known-severity vulnerability. It was published on August 13, 2026.
Technical 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.
Remediation
Check the references section for vendor advisories, patches, and mitigation guidance. If immediate patching is not possible, review the CVSS vector to understand the attack surface and apply compensating controls such as network segmentation or access restrictions.
Frequently Asked Questions
What is CVE-2026-73557?
CVE-2026-73557 is a known-severity vulnerability. It was published on August 13, 2026.
How severe is CVE-2026-73557?
CVSS score information is not yet available for this vulnerability. Check back as the CVE record is updated by NVD analysts.
How do I fix or remediate CVE-2026-73557?
Check the references section for vendor advisories, patches, and mitigation guidance. If immediate patching is not possible, review the CVSS vector to understand the attack surface and apply compensating controls such as network segmentation or access restrictions.
How can CyberStrike help with CVE-2026-73557?
CyberStrike's AI-powered security agents can automatically detect CVE-2026-73557 across your infrastructure using autonomous pentesting, DAST scanning, and HackBrowser. The platform continuously monitors for known vulnerabilities and provides actionable remediation guidance prioritized by real-world exploitability.
How CyberStrike Helps
AI agents map your attack surface to find vulnerabilities like this one.
Automated penetration testing that runs continuously, not just quarterly.
AI-driven PR review catches vulnerable dependencies before they ship.
Browser-based exploitation validates findings with real proof-of-concept.