Vulnerability in vllm (CVE-2026-105758)
Summary
Vulnerability in vllm (CVE-2026-105758). Risk of unauthorized operations or information disclosure. Exploitable via POST /tokenize. Mitigation: upgrade to 0.30.0 or later.
Original description
The description as published by the source.
Summary
An unauthenticated remote attacker can exhaust the memory of the vLLM API-server process by raising the request-level media_io_kwargs.video.max_frames and fps knobs on any deployment serving a Qwen2-VL or Qwen3-VL model. 74 extra bytes of JSON took the server's peak RSS from 2 271 MiB to 13 629 MiB over unauthenticated POST /tokenize.
The num_frames ceiling reported in GHSA-vxqj-p4gw-9h4c and fixed by open PR #51969 does not reach this path: the Qwen samplers do not read num_frames at all. The same knobs were already capped upstream for GLMGAVideoBackend as an accepted security fix in 8b6de0eb9 (PR #54935, merged 2026-09-04); that cap never reached Qwen.
Details
Relationship to GHSA-vxqj-p4gw-9h4c and PR #51969 (read this first)
GHSA-vxqj-p4gw-9h4c reported that request-level media_io_kwargs.video.num_frames overrides the engine frame-count ceiling, and open PR #51969 fixes it by clamping num_frames inside VideoMediaIO.merge_kwargs.
That clamp does not reach the Qwen samplers. Qwen2VLVideoBackend and Qwen3VLVideoBackend do not read num_frames at all — Qwen2VLVideoBackend's own docstring says so ("num_frames is ignored (fps-driven, like the Qwen3-VL loader)"). They bound on max_frames, read from the same merged dict with no ceiling:
# vllm/multimodal/video.py — Qwen3VLVideoBackend.compute_frames_index_to_sample
min_frames = kwargs.get("min_frames", 4)
max_frames = kwargs.get("max_frames", 768)
num_frames =…
Source: NIST National Vulnerability Database NVD GitHub Advisory
Auto-generated summary
Assembled by SecPulse from the source data (CVSS, CWE, affected versions) using templates — no AI involved. Check the original description and references below for exact details.
A vulnerability tracked as CVE-2026-105758 has been found in vllm.
Attackers can target a specific entry point like POST /tokenize over the network to misuse the product.
Risk of unauthorized operations or information disclosure. CVSS score: 5.3/10.
What to do: upgrade vllm to 0.30.0 or later.
If unsure, ask your IT team or search "vllm CVE-2026-105758" on the vendor's site.
CVE-2026-105758 (vllm) — CWE-770 / CVSS v3 5.3
Attack vector: remote (network-reachable) / unauthenticated / no user interaction
Attack surface: POST /tokenize / media_io_kwargs.video.max_frames / fps / num_frames
Patched: 0.30.0 — apply immediately
Plan: 1) Audit SBOM/dependencies, 2) Stage→prod upgrade, 3) Add WAF/proxy monitoring on affected endpoints, 4) Hunt IOCs in logs.
Refs: see the GHSA / vendor advisory / patched release linked on this page.
A vulnerability (CWE-770) exists in vllm. Attackers reach the vulnerable code path via POST /tokenize without authentication.
Check the vendor advisory for the affected versions of vllm. Attack surface: POST /tokenize / media_io_kwargs.video.max_frames / fps / num_frames.
Severity: Medium (CVSS 5.3/10). Risk of unauthorized operations or information disclosure.
Update to 0.30.0.
Until the patch is applied: disable the affected feature, apply WAF rules, or restrict access via network ACLs.
Search webserver/proxy logs for unusual POST /tokenize requests with malformed payloads or SQL meta-characters. Run grep -r 'vllm' . against your dependency files (package-lock.json, requirements.txt, go.sum) to find affected services.
Response actions
Concrete steps and command examples for SOC/SRE teams to execute in order
-
1Identify exposure identify
grep -r 'vllm' . | grep -v node_modulesリポジトリと本番環境の依存ファイル (package-lock.json / requirements.txt / go.sum / Gemfile.lock 等) で
vllmを grep し、稼働しているサービス・バージョンを把握する。 -
2Hunt for indicators of compromise detect
grep 'POST /tokenize' /var/log/nginx/access.log | grep -E '(unusual_payload|sqli_pattern)'アクセスログで
POST /tokenizeへの異常なリクエスト (不正な認証ヘッダ・SQLメタ文字)を過去 30〜90日分捜索。WAF/SIEM があれば該当パスのアラート発火履歴を確認。 -
3Apply patch patch
Upgrade vllm to 0.30.0ステージング環境で 0.30.0 に上げて回帰テスト → 本番反映。回帰テストはアプリの主要ハッピーパスと、Step 2 で見つけた異常検知の続報チェックを含めること。
-
4Post-deployment verification verify
Replay attack against POST /tokenize on staging to confirm patch closes the vectorパッチ適用後、ステージングで PoC または同等の悪用パターンを再現して脆弱性が閉じたことを確認。本番では Step 2 と同じログクエリでアラート再発が無いか継続監視。