CVE-2026-54234: vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then converted to negative one when the engine selects the next live token for a request and is written back into the drafter's input ids; that out-of-vocabulary value is later consumed by the model's embedding and attention path and crashes the engine worker with a GPU device-side assertion. The same triggering request sequence is reachable through the public gRPC Generate and Abort endpoints, so a remote client that can send generation requests can crash the shared engine worker, aborting concurrent requests and causing a service-wide denial of service for other clients of the deployment until the worker is restarted. This issue is fixed in version 0.24.0.

Description from the National Vulnerability Database, published by GitHub

What it means for a small business

If your business uses vllm, install the vendor's update. If you're not sure whether you use it, ask whoever manages your computers.

How likely is an attack? The Exploit Prediction Scoring System, run by FIRST (the global association of security response teams), puts the odds of this one being exploited in the next 30 days at 0.62%, higher than 49% of CVEs. That’s low, but low isn’t zero, and the odds jump the day someone publishes an exploit. Scored October 29, 2026, updated daily.

Check your own setup

Website check: reads what your site runs from the outside (WordPress version, plugins it can see, old server software) and matches it against this database.
Server check: what your web server says about itself, and whether that version has known flaws.

Sources

Part of the Huttenlocher Technologies vulnerability database, refreshed twice a day from CISA and the NVD. The title, category and plain-language guidance are ours; check the linked record before acting on a detail. This is general guidance, not a review of your systems.

Weekly security brief

The flaws that hit local websites, every Friday.

One short email: what got exploited this week, which WordPress plugins broke, and what to check. Unsubscribe in one click.