CVE-2026-100647: Improper Input Validation in vllm-project vllm
vLLM versions before 0.29.0 contain a denial-of-service vulnerability in the cache_salt parameter accepted on OpenAI-compatible and Anthropic API endpoints, which lacks maximum length validation and is processed on the single EngineCore scheduler thread. Unauthenticated attackers can send HTTP requests with multi-hundred-megabyte salt values that trigger expensive pickle serialization and SHA-256 hashing, stalling the scheduler thread and denying service to all concurrent requests.
AI Analysis
Technical Summary
CVE-2026-100647 describes a denial-of-service vulnerability in vLLM versions prior to 0.29.0. The vulnerability arises because the cache_salt parameter accepted by OpenAI-compatible and Anthropic API endpoints does not enforce a maximum length limit. This parameter is processed on the single EngineCore scheduler thread. An attacker can send HTTP requests with multi-hundred-megabyte salt values, triggering costly pickle serialization and SHA-256 hashing operations. These expensive computations stall the scheduler thread, resulting in denial of service for all concurrent requests.
Potential Impact
An unauthenticated attacker can cause a denial-of-service condition by sending specially crafted requests with very large cache_salt values. This stalls the scheduler thread responsible for processing requests, effectively blocking all concurrent requests and disrupting service availability.
Mitigation Recommendations
A fix is available in vLLM version 0.29.0 and later. Users should upgrade to version 0.29.0 or newer to remediate this vulnerability. No additional mitigation guidance is provided.
CVE-2026-100647: Improper Input Validation in vllm-project vllm
Description
vLLM versions before 0.29.0 contain a denial-of-service vulnerability in the cache_salt parameter accepted on OpenAI-compatible and Anthropic API endpoints, which lacks maximum length validation and is processed on the single EngineCore scheduler thread. Unauthenticated attackers can send HTTP requests with multi-hundred-megabyte salt values that trigger expensive pickle serialization and SHA-256 hashing, stalling the scheduler thread and denying service to all concurrent requests.
CVSS v4.0
Score 6.9medium
Affected software
vllm-project
vllm
pkg:github/vllm-project/vllmRun on your own infrastructure? Check whether these packages are installed with threat-finder — our free open-source scanner.
AI-Powered Analysis
Machine-generated threat intelligence
Technical Analysis
CVE-2026-100647 describes a denial-of-service vulnerability in vLLM versions prior to 0.29.0. The vulnerability arises because the cache_salt parameter accepted by OpenAI-compatible and Anthropic API endpoints does not enforce a maximum length limit. This parameter is processed on the single EngineCore scheduler thread. An attacker can send HTTP requests with multi-hundred-megabyte salt values, triggering costly pickle serialization and SHA-256 hashing operations. These expensive computations stall the scheduler thread, resulting in denial of service for all concurrent requests.
Potential Impact
An unauthenticated attacker can cause a denial-of-service condition by sending specially crafted requests with very large cache_salt values. This stalls the scheduler thread responsible for processing requests, effectively blocking all concurrent requests and disrupting service availability.
Mitigation Recommendations
A fix is available in vLLM version 0.29.0 and later. Users should upgrade to version 0.29.0 or newer to remediate this vulnerability. No additional mitigation guidance is provided.
Technical Details
- Data Version
- 5.2
- Assigner Short Name
- VulnCheck
- Date Reserved
- 2026-09-26T02:33:07.898Z
- Cvss Version
- 4.0
- State
- PUBLISHED
Threat ID: 6ab7c9a7f7a7c5410652fd2f
Added to database: 09/26/2026, 13:33:27 UTC
Last enriched: 09/26/2026, 14:03:36 UTC
Last updated: 09/27/2026, 04:31:30 UTC
Views: 13
Community Reviews
0 reviewsCrowdsource mitigation strategies, share intel context, and vote on the most helpful responses. Sign in to add your voice and help keep defenders ahead.
Want to contribute mitigation steps or threat intel context? Sign in or create an account to join the community discussion.
Actions
Updates to AI analysis require Pro Console access. Upgrade inside Console → Billing.
Need more coverage?
Upgrade to Pro Console for AI refresh and higher limits.
For incident response and remediation, OffSeq services can help resolve threats faster.
Latest Threats
Check if your credentials are on the dark web
Instant breach scanning across billions of leaked records. Free tier available.