vLLM Denial of Service via Unvalidated mm_processor_kwargs
vLLM
An unvalidated parameter vulnerability in vLLM allows remote clients to exhaust server memory using large multimodal processor configurations. The issue can cause the shared inference server to crash and terminate.
disclosed CVENo confirmed exploitation, so this record is tracked as a disclosed vulnerability and not counted as an incident. It moves into the counts if CISA or the vendor confirms exploitation. Why
- fix status
- Fix status unknown
- first report
- 21 Oct 2026
- last update
- 21 Oct 2026
Timeline · 1
Sources · 1
Advisories and databases
- 2026-10-21GHSA-3cgq-c484-rx5j: Duplicate Advisory: vLLM 0.7.3 before 0.31.0 DoS via Unvalidated mm_processor_kwargsGitHub Advisory Database · github.com/advisories/GHSA-3cgq-c484-rx5j
Titles and links belong to their publishers. The summary above is written by gcdRadar in its own words; data is CC BY-SA 4.0.
Classification
- lane
- Patch now
- AI role
- Attack on AI
- categories
- Infrastructure vulnerability
- severity
- 2 · Disclosed vulnerability
- fix status
- Fix status unknown
- targets
- Framework / tooling
- owasp llm top 10
- LLM03:2025 Supply Chain
- mitre atlas
- no technique for these categories
- dates
- happened 21 Oct 2026 · first reported 21 Oct 2026
- origin
- GitHub Advisory DatabaseVulnerability record from NVD / the GitHub Advisory Database. This product uses the NVD API but is not endorsed or certified by the NVD.
- record
- confidence 0.95 · created 2026-10-23 22:00Z · updated 2026-10-23 22:00Z
Lane, AI role, categories and severity come from the classifier; the OWASP and ATLAS rows are derived from the categories. How these are assigned
Copy a brief
This record as a short Markdown brief, for a ticket, a chat or a coding agent.
the brief as text
# vLLM Denial of Service via Unvalidated mm_processor_kwargs https://gcdradar.vercel.app/incidents/vllm-denial-of-service-via-unvalidated-mm-processor-kwargs · AIR-2026-1485 · closed 2026-10-21 - Lane: Patch now - Severity: 2 (Disclosed vulnerability) - Fix status: Fix status unknown - First reported: 2026-10-21 - AI role: Attack on AI - Products: vLLM ## Summary An unvalidated parameter vulnerability in vLLM allows remote clients to exhaust server memory using large multimodal processor configurations. The issue can cause the shared inference server to crash and terminate. ## What to do No action recorded; read the sources below and the vendor's advisory. ## Latest updates - 2026-10-21: GHSA-3cgq-c484-rx5j: Duplicate Advisory: vLLM 0.7.3 before 0.31.0 DoS via Unvalidated mm_processor_kwargs (https://github.com/advisories/GHSA-3cgq-c484-rx5j) ## Sources - 2026-10-21 GitHub Advisory Database: GHSA-3cgq-c484-rx5j: Duplicate Advisory: vLLM 0.7.3 before 0.31.0 DoS via Unvalidated mm_processor_kwargs https://github.com/advisories/GHSA-3cgq-c484-rx5j Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.
Related
- closed6.9vLLM CVE-2026-126395: vLLM information disclosure vulnerability in sanitize_message()
Patch now · Disclosed · Patched · since 13 Nov 2026 · CVE-2026-126395
AI - closed7.1vLLM CVE-2026-126399: vLLM resource exhaustion vulnerability via video frame limits bypass
Patch now · Disclosed · Patched · since 13 Nov 2026 · CVE-2026-126399
AI - closed8.2vLLM CVE-2026-126398: vLLM reachable assertion vulnerability in GPUModelRunner._init_xdrope_positions
Patch now · Disclosed · Patched · since 13 Nov 2026 · CVE-2026-126398
AI - closed6.9vLLM CVE-2026-126394: vLLM resource exhaustion vulnerability in OpenCVDynamicOpenPanguVideoBackend
Patch now · Disclosed · Patched · since 13 Nov 2026 · CVE-2026-126394
AI - closed8.7vLLM CVE-2026-126397: vLLM quadratic CPU work vulnerability in Cohere2 Vision multimodal preprocessing
Patch now · Disclosed · Patched · since 13 Nov 2026 · CVE-2026-126397
AI - closed6.9vLLM CVE-2026-126396: vLLM resource exhaustion vulnerability in MiniMax M3 VL video processor
Patch now · Disclosed · Patched · since 13 Nov 2026 · CVE-2026-126396
AI