CVE-2026-115916VulnerabilityPublished 26 Oct 2026Updated 29 Oct 2026

Prior to 0.32.0, vllm/tokenizers/hf.py computes max_token_id with an off-by-one upper bound, allowing a pre-tokenized

Affects vllmSource GCVE Database

Compare

Authoritative record:CVE-2026-115916· Medium 6.5

Not known to be exploited · CVSS 6.5 Medium : upgrade to 0.32.0 when convenient.

Data as of 29 Oct 2026, 14:41 UTC

CVSS 3.1 base score

6.5Medium

How severe the flaw is on the 0–10 CVSS scale.

Thin evidence

One scorer or none

CVSS scores by source
CNA (GitHub_M)	CVSS 3.1	6.5

EPSS: 0.3% chance of exploitation in 30 days – higher than 24% of CVEsExploit Prediction Scoring System: the likelihood attackers will use this flaw soon.
Patch: Available – fixed in 0.32.0

CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H

Attack Vector: Network
Attack Complexity: Low
Privileges Required: Low
User Interaction: None
Scope: Unchanged
Confidentiality: None
Integrity: None
Availability: High

Bold values are the worst case: easiest to exploit or largest impact.

Attack Vector: Network: The vulnerable component is bound to the network stack and the set of possible attackers extends beyond the other options listed, up to and including the entire Internet. Such a vulnerability is often termed 'remotely exploitable' and can be thought of as an attack being exploitable at the protocol level one or more network hops away (e.g., across one or more routers).
Attack Complexity: Low: Specialized access conditions or extenuating circumstances do not exist. An attacker can expect repeatable success against the vulnerable component.
Privileges Required: Low: The attacker is authorized with (i.e., requires) privileges that provide basic user capabilities that could normally affect only settings and files owned by a user. Alternatively, an attacker with Low privileges may have the ability to cause an impact only to non-sensitive resources.
User Interaction: None: The vulnerable system can be exploited without any interaction from any user.
Scope: Unchanged: An exploited vulnerability can only affect resources managed by the same security authority. In this case, the vulnerable component and the impacted component are either the same, or both are managed by the same security authority.
Confidentiality: None: There is no loss of confidentiality within the impacted component.
Integrity: None: There is no loss of integrity within the impacted component.
Availability: High: There is total loss of availability, resulting in the attacker being able to fully deny access to resources in the impacted component; this loss is either sustained (while the attacker continues to deliver the attack) or persistent (the condition persists even after the attack has completed). Alternatively, the attacker has the ability to deny some availability, but the loss of availability presents a direct, serious consequence to the impacted component (e.g., the attacker cannot disrupt existing connections, but can prevent new connections; the attacker can repeatedly exploit a vulnerability that, in each instance of a successful attack, leaks a only small amount of memory, but after repeated exploitation causes a service to become completely unavailable).

Definitions: FIRST CVSS v3.1 specification.

Exploitation evidence

Disclosed: evidence found.
CVE published 26 Oct 2026
Detectable: no evidence on record.
No public nuclei template indexed
Weaponized: no evidence on record.
No public exploit code indexed in Metasploit or Exploit-DB
Exploited in the wildExploited: no evidence on record.
Not in CISA KEV and no source reports exploitation
Observed by OffSeq MirageSeen by Mirage: no evidence on record.
No attack traffic recorded by OffSeq Mirage sensors

A missing step is not evidence of safety. Sources: cve.org (opens in a new tab), the Metasploit, Exploit-DB and nuclei-templates index as of 29 Oct 2026.

Disclosed: The CVE record is public on cve.org. Its publication is t = 0.
Detectable: A public nuclei template can find exposed instances, for defenders and attackers alike.
Weaponized: A Metasploit exploit module or Exploit-DB entry exists (Metasploit scanner modules do not count), or CISA SSVC or the CVSS exploit maturity reports a public proof of concept.
Exploited in the wild: CISA KEV lists it, CISA SSVC rates exploitation active, or a source reports attacks.
Observed by OffSeq Mirage: OffSeq Mirage honeypot sensors recorded attack traffic for this CVE, or found exposed hosts without attack traffic.

Moments are spaced evenly, not to scale; a dashed line marks a gap of six months or more. Deltas count from t = 0. Sources that give only a date (CISA KEV, Exploit-DB, or a time of exactly 00:00Z) read “same day” rather than hours. Metasploit modules carry the flaw’s disclosure date, not their release date, and nuclei templates carry no date, so both appear above but not on the timeline.

What to do

Upgrade vllm to 0.32.0.

Patches and advisories

Affected software

Product	Affected	Fixed in
PyPIvllm`pkg:pypi/vllm`Other threats	Affected <0.32.0	Fixed in 0.32.0

Check your own hosts for these packages with threat-finder (opens in a new tab), our free open-source scanner.

AI-suggested

Upgrade to vLLM version 0.32.0 or later, where this off-by-one error and related assertion failure have been fixed. The vendor has provided an official patch and advisory. No other mitigation is required once the update is applied.

Description

vLLM versions prior to 0.32.0 contain an off-by-one error in the computation of max_token_id in the hf.py tokenizer module. This flaw allows a pre-tokenized prompt ID equal to the vocabulary size to bypass validation. Under specific conditions involving tensor-parallel size and vocabulary size, this leads to a CUDA device-side assertion failure that terminates the engine. The issue is fixed in version 0.32.0.

Impact

The vulnerability causes a denial of service by triggering a CUDA device-side assertion failure that terminates the vLLM engine. There is no impact on confidentiality or integrity. The disruption affects availability of the service running the vulnerable vLLM versions.

Summary

vLLM, an inference and serving engine for large language models, had a vulnerability in versions before 0.32.0 where the max_token_id was computed with an off-by-one upper bound in the vllm/tokenizers/hf.py file. This allowed a pre-tokenized prompt ID equal to the vocabulary size to pass validation incorrectly. When the tensor-parallel size is one and the vocabulary size is an exact multiple of 64, the VocabParallelEmbedding.forward function does not mask this out-of-bound value before it is used in F.embedding. This causes a CUDA device-side assertion failure, resulting in the termination of the engine. The vulnerability has a CVSS 3.1 base score of 6.5 (medium severity) with no confidentiality or integrity impact but a high availability impact. The issue is resolved in vLLM version 0.32.0.

References

Community reviews

Loading reviews…