Improper Handling of Exceptional Conditions Affecting vllm package, versions [,0.30.0)
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 applications- Snyk IDSNYK-PYTHON-VLLM-20877481
- published13 Oct 2026
- disclosed5 Oct 2026
- creditUnknown
Introduced: 5 Oct 2026
NewCVE-2026-105757 (opens in a new tab)How to fix?
Upgrade vllm to version 0.30.0 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 Improper Handling of Exceptional Conditions via validate_tokens in the structured output backend, when a request carrying invalid structured-output grammar causes the engine to process speculative-decoding padding sentinels (-1 values) as real token IDs. An authenticated remote attacker can send a malformed structured-output request that passes -1 sentinel tokens into the grammar validator, triggering an unhandled error that stops the entire inference engine and denies service to all users.