VulnSocial

CVE-2026-93989Disclosure(vllm / vllm)

LOWCVSS 4.3 · MEDIUM

Signal is active with 1 mentions in latest observed window

Immediate actions

  • Patch vllm vllm systems immediately

Recommended action window: Monitor and triage in normal cycle

NVD description

vLLM through 0.29.0 fails to properly validate bad_words token indices against the model's generation output width in SamplingParams.update_from_tokenizer(). Attackers can supply out-of-bounds token indices that corrupt logits memory of concurrent requests, causing different in-flight HTTP requests to return incorrect tokens.

vllm CVEsBack to dashboard
0.5/ 10 priority

Sources & remediation

Patch / fix
Vendor / third-party advisories
Other references
Weakness type (CWE)
CWE-129

Priority

LOW

Exploitation

NONE

PoC

NONE

Patch

AVAILABLE

Momentum

STABLE

Are you affected?

If you run products in this scope, you should treat this CVE as relevant to your environment.

  • vllm

Threat summary

  • Patch or workaround signal is available
  • 2 mentions across 2 observed days
  • Momentum state: stable

What's happening

  • Patch or workaround mentioned in 1 signal
  • Technical details provided in 2 signals
  • Disclosure: 1 classified signal
  • Peaked 1d ago at 1 mentions (2026-09-20); latest day: 1
  • 2 total mentions across 2 days

Affected systems

Vendors
vllm
Products
vllm

Deep dive

Activity timeline2 mentions / 2d
00111Mentions · 2026-09-20: 1Mentions · 2026-10-28: 1Patch / Workaround · 2026-10-28: 1Technical Details · 2026-09-20: 1Technical Details · 2026-10-28: 109-2010-28
Signal classification2 categories
Disclosure
150.0%
Patch
150.0%
Referenced assets2 URLs
Classification over time
DateTotalLabels
2026-09-201
Disclosure1
2026-10-281
Patch1
Full discourse2 posts
  • Hugo | DevOps | Cybersecurity@HugoValters
    Patch

    CVE-2026-93989 vLLM out-of-bounds access lets crafted bad_words indices corrupt logits, so concurrent requests return wrong tokens. CVSS 3.1. Patch available, update now. #CVE #infosec #vLLM https://www.valtersit.com/cve/CVE-2026-93989 #vLLM #CVEALERT #CVE #infosec #Linux #XSS #valtersit #snippet #SysAdmin #cybersecurity #devsecops #devops #developer #100daysofcode #git #github #gitlab #redteam #blueteam #ethicalhacker #cybersecurityawareness #cybersecuritynews #cybersecuritytips #python #hacker #kali #ubuntu #debian #docker

    Post summary

    The post discloses CVE-2026-93989 in vLLM, describing an out-of-bounds access that can corrupt logits and cause wrong tokens, and states that a patch is available with an update now call.

    0000043
    1.1K followersView on X
  • CVE@CVEnew
    Disclosure

    CVE-2026-93989 vLLM through 0.29.0 fails to properly validate bad_words token indices against the model's generation output width in SamplingParams.update_from_tokenizer(). Attacker… https://www.cve.org/CVERecord?id=CVE-2026-93989

    Post summary

    The text discloses CVE-2026-93989, detailing an improper validation flaw in vLLM versions up to 0.29.0 within the SamplingParams.update_from_tokenizer() method, without mentioning remediation or exploitation evidence.

    00000886
    58.1K followersView on X
CPE platform detail1 entries

1 of 1 entries

PartVendorProductVersionTarget SWTarget HW
Appvllmvllm---

Explore more

Back to dashboardvllm CVEsvllm vulnerabilities