Patch Forecast
Yellow warning
Plan into this sprint.

CVE-2026-100647

Vllm: improper input validation

Risk score
56/100
CVSS
7.5 v3.1
Exploit chance (30 days)
0.3%
Published
26 Sep 2026

Open in the live forecast Today's forecast

What to know

What it is
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-hund…
Affected
Vllm before 0.29.0
Fix
Upgrade Vllm to 0.29.0 or later.
Exploitation
No known exploitation

Story map

Story map: VllmTHE FLAWSTHE FIXCVE-2026-100647Improper input validationCVSS 7.5 · EPSS 0%CVE-2026-100652Improper input validationCVSS 7.5 · EPSS 0%CVE-2026-105759VLLM is an inference andserving engine for largelanguage modelsCVSS 7.5 · EPSS 0%CVE-2026-103241Denial of serviceCVSS 7.5 · EPSS 1%Upgrade Vllm0.29.0 · 0.30.0or later on your release lineWho is exposedVllmAMBER · PATCH THIS WEEKVllmvLLM versions 0.22.0 through0.23.0 fail to validatestop_token_ids againstvocabulary bounds in Rust HTTPand gRPC frontends, allowingout-of-vocabulary token IDs toreach…Where it standsPublic proof of conceptCONNECTS TOSame weaknessDenial of service (CWE-400)253 other warnings in 30 days →Same productEarlier Vllm warnings15 in the last 90 days →
Story map: the flaws, the fix, how it unfolded and what it connects to. Boxes link to the record or article.

At a glance

Vllm: Amber, patch this week. vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA ten…

The flaws

The fix

Upgrade Vllm: 0.29.0 · 0.30.0 or later on your release line

Who is exposed: Vllm

Where it stands

  • Public proof of concept

Connects to

Why it is Yellow

  • CVSS 7.5 high
  • Remotely exploitable without logging in

References

Original 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.

Last activity 26 Sep 2026. Patch Forecast updates every hour.