Allocation of Resources Without Limits or Throttling Affecting vllm package, versions [,0.32.0)
Threat Intelligence
The probability is the direct output of the EPSS model, and conveys an overall sense of the threat of exploitation in the wild. The percentile measures the EPSS probability relative to all known EPSS scores. Note: This data is updated daily, relying on the latest available EPSS model version. Check out the EPSS documentation for more details.
Do your applications use this vulnerable package?
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applicationsSnyk Learn
Learn about Allocation of Resources Without Limits or Throttling vulnerabilities in an interactive lesson.
Start learning- Snyk IDSNYK-PYTHON-VLLM-21755393
- published27 Oct 2026
- disclosed26 Oct 2026
- creditC1ves
Introduced: 26 Oct 2026
NewCVE-2026-115913 (opens in a new tab)How to fix?
Upgrade vllm to version 0.32.0 or higher.
Overview
vllm is an A high-throughput and memory-efficient inference and serving engine for LLMs
Affected versions of this package are vulnerable to Allocation of Resources Without Limits or Throttling via the MultiHeadedAttention module in phi4mm_utils.py, where attention matrices computed during Phi-4 multimodal audio processing are retained as instance attributes (self.attn) on the module object. This causes unbounded memory growth across inference requests, as each forward pass overwrites the attribute with a new tensor that is never released, allowing an attacker who can submit audio inference requests to exhaust available memory and crash the server.