CVE-2026-94623: vLLM through 0.29.0 Denial of Service via NIXL Multi-Prompt Assertion Failure
vLLM through 0.29.0 contains a denial of service vulnerability in the NIXL connector's prefix caching implementation that fails to properly validate block counts across multi-prompt completion requests in prefill/decode disaggregated deployments. Attackers can trigger an assertion failure in NixlBaseConnectorWorker.applyprefixcaching by submitting completion requests with multiple prompts of varying lengths, causing the decode worker to terminate and become unavailable until restarted.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
vLLM (NIXL connector / NixlBaseConnectorWorker._apply_prefix_caching)to a version that resolves this vulnerability.Fixed in 0.29.0
Event History
Frequently Asked Questions
Which deployments are affected?
The issue affects vLLM deployments using the NIXL connector with prefix caching in a prefill/decode disaggregated configuration. The affected versions are through 0.29.0.
What does an attacker need to do to trigger the failure?
An attacker can submit a completion request containing multiple prompts with varying lengths. No privileges or user interaction are required according to the provided vector.
What is the operational impact?
The malformed multi-prompt request can trigger an assertion failure in NixlBaseConnectorWorker._apply_prefix_caching. This terminates the decode worker, making it unavailable until it is restarted.