CVE-2026-115914
Medium · 6.5CVSS
6.5
EPSS
0.4%
Published
Oct 26, 2026
CISA due date
—
The record
vLLM is an inference and serving engine for large language models. From 0.11.1 until 0.32.0, the DeepSeek-OCR, DeepSeek-OCR-2, and Unlimited-OCR image path allows a remote client that can reach the OpenAI-compatible API to submit a legal image whose normalized pixels sum to zero when that image is alone in its per-step encoder batch. _parse_and_validate_image_input returns no embedding, embed_multimodal propagates that value, and sanity_check_mm_encoder_outputs triggers a fatal assertion that shuts down the engine for all users. This issue is fixed in version 0.32.0.
Affected packages
| Ecosystem | Package | Introduced | Fixed |
|---|
Related vulnerabilities
- CVE-2026-119339 Tesseract is an open source OCR engine. CWE-617
- CVE-2026-119139 Wasmtime 46.0.0 before 48.0.6 and 49.0.0 before 49.0.3 contains a leftover assertion in… CWE-617
- CVE-2026-119142 Wasmtime 46.0.0 before 48.0.6 and 49.0.0 before 49.0.3 contains a denial of service… CWE-617
- CVE-2026-119138 Wasmtime 46.0.0 before 48.0.6 and 49.0.0 before 49.0.3 contains a denial of service… CWE-617
- CVE-2026-119135 Wasmtime 46.0.0 before 48.0.6 and 49.0.0 before 49.0.3 panics when the host reads a… CWE-617
- CVE-2026-119132 Wasmtime from 49.0.0 before 49.0.3 contains a denial-of-service vulnerability that allows… CWE-617
- CVE-2026-116467 A security flaw has been discovered in ggml-org llama.cpp up to 0.7.0. CWE-617
- CVE-2026-102483 Reachable exit in NTPsec ntpd 1.1.4 through 1.2.5 allows a remote attacker to terminate… CWE-617