CVE-2026-111392Medium· 4.8▾ TwilightPoC availableSGLang through 0.5.21 trusts caller-supplied mm_hashes values as the multimodal prefix-cache key, allowing remote attackers to poison cached KV entries across requests. Attackers can send a different image under a known or predicted hash…
▾ Twilight zone — High severity, or a signal on a lesser flaw
impact 26.4 · likelihood 0 · exploitation 12
A public proof-of-concept already exists for this vulnerability — see Exploit availability below.
Stakeholder-Specific Vulnerability Categorization from CISA's ADP record at CVE.org: whether exploitation is observed, whether an attack can be automated, and how much of the system is at stake.
Exploit-prediction probability, daily snapshots since Oct 17.
Disclosure to exploitation, from the record and what we observed since indexing it.
Disclosed via NVD
Exploit / PoC code exists
0.2%
SGLang through 0.5.21 trusts caller-supplied mm_hashes values as the multimodal prefix-cache key, allowing remote attackers to poison cached KV entries across requests. Attackers can send a different image under a known or predicted hash via /generate to disclose another request's image content or corrupt victims' responses.
Refer to the linked advisories for vendor-supplied fixes and affected version ranges.
Field changes observed since this record was first indexed.
Connected by shared product, vendor, weakness, or advisory.
CVE-2026-112267Medium· 6.5SGLang through 0.5.22 SSRF via bootstrap_host and bootstrap_port on Decode Worker
CVE-2026-112269Medium· 6.5SGLang through 0.5.22 Memory Exhaustion via Unauthenticated Mooncake Decode Registrations
CVE-2026-112266Medium· 5.9SGLang through 0.5.22 Cross-Request Abort via rid Prefix Matching
CVE-2026-112082High· 7.5sgl-project SGLang Expert-Distribution Recording Endpoint expert_distribution.py Scheduler.expert_distribution_handle denial of service
CVE-2026-112083Medium· 5.3sgl-project SGLang Weight Exporter weight_exporter.py WeightExporter.send_weights_to_remote_instance denial of service
CVE-2026-111391High· 8.1SGLang 0.5.10 through 0.5.22 contains a code injection vulnerability in _evaluate_filter() that allows remote attackers to execute Python code by escaping the eval() sandbox through re.search __globals__