HighVulnerabilityLLM-specific
CVE-2026-73557: vLLM race condition bypasses sparse tensor validation in prompt_embeds
- Source
- NVD(opens in a new tab)
- Identifier
- CVE-2026-73557
- Published
- Record updated
- Affected
- vllm >= 0.21.0, < 0.26.0
- Fixed in
- 0.26.0
- released July 25, 2026, 19 days before disclosure
- Known exploitation
- Not listed in the CISA Known Exploited Vulnerabilities catalog at the last check.
- EPSS
- 0.2%
Summary
vLLM, an inference and serving engine for large language models, contains a flaw from 0.20.2rc0 through 0.26.0 in safe_load_prompt_embeds in vllm/renderers/embed_utils.py. The function toggles the process-global torch.sparse.check_sparse_tensor_invariants setting, and concurrent prompt_embeds parts submitted to POST /v1/chat/completions can race that state, letting an invalid sparse tensor reach tensor.to_dense despite the CVE-2025-62164 guard when enable_prompt_embeds is enabled.
Mitigation
This issue is fixed in version 0.26.0.
Affected packages in the Exposure Registry
Matched by package name and ecosystem. Each entry shows whether the package delegates to a language model and how many tracked packages depend on it.
- vllmPyPILLM dependency since 2023-08-25 · 15 tracked dependents
Topics
Related items
- MediumCVE-2026-126399: vLLM resource exhaustion via video frame limit bypass in chat completionsSimilar attack · NVD/CVE Database
- MediumCVE-2026-126398: vLLM reachable assertion crash in GPUModelRunner via embeddings-only requestsSimilar attack · NVD/CVE Database
- HighCVE-2026-126397: vLLM denial of service via unbounded max_patches in multimodal preprocessingSimilar attack · NVD/CVE Database
- MediumCVE-2026-126396: vLLM resource exhaustion in MiniMax M3 VL video processor via min_pixelsSimilar attack · NVD/CVE Database
- MediumCVE-2026-126394: vLLM resource exhaustion in OpenCVDynamicOpenPanguVideoBackend via num_framesSimilar attack · NVD/CVE Database