EUVD-2026-69410
GHSA-rqxm-jx33-wrqr
Severity by source
AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
Network-reachable endpoint with no auth or interaction required; impact is availability only via crash; no scope change beyond the llama.cpp process.
Primary rating from Vendor (mitre).
CVSS VectorVendor: mitre
Lifecycle Timeline
5DescriptionCVE.org
llama.cpp through commit 97f06e9, when started with the --reranking flag, allows remote attackers to cause a denial of service (std::bad_alloc and HTTP 500) via a negative top_n value in a POST request to /rerank.
AnalysisAI
Remote unauthenticated denial of service in llama.cpp (through commit 97f06e9) allows any network attacker to crash the inference server by submitting a POST request to the /rerank endpoint with a negative top_n value, triggering a std::bad_alloc exception and an HTTP 500 response. The attack surface is exclusively present when the server is started with the non-default --reranking flag. …
Unlock full vulnerability intelligence
- Risk assessment & exploitation conditions
- Attack chain visualization
- Remediation with exact patch versions
- Threat intelligence from 22 sources
- Personal watchlist & email alerts
Free forever · No credit card required
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata
Vulnerability AssessmentAI
| Exploitation | The --reranking flag must be explicitly passed at llama.cpp server startup - this is a non-default operator choice. … Additional conditions and limiting factors are described in the full assessment. |
| Risk Assessment | The CVSS 7.5 score (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H) accurately reflects the exploitability characteristics once the --reranking mode is active: a single crafted HTTP request, no authentication, no complexity barrier. … Full risk analysis with EPSS, KEV, and SSVC signal comparison available after sign-in. |
| Exploit Scenario | Full exploit scenario with step-by-step reproduction available after sign-in. |
| Remediation | No explicitly patched version or commit is confirmed in available data; operators should monitor the ggml-org/llama.cpp GitHub repository (https://github.com/ggml-org/llama.cpp) for a fix that adds bounds validation on the top_n parameter. … Detailed patch versions, workarounds, and compensating controls in full report. |
Recommended ActionAI
Within 24 hours: identify all llama.cpp deployments in production and audit which ones are running with the --reranking flag enabled. …
Sign in for detailed remediation steps and compensating controls.
Threat intelligence, references, and detailed analysis are available after sign-in.
Same weakness CWE-674 – Uncontrolled Recursion
View allSame technique Denial Of Service
View allVendor StatusVendor
SUSE
Severity: Important| Product | Status |
|---|---|
| openSUSE Tumbleweed | Fixed |
Share
External POC / Exploit Code
Leaving vuln.today