6.8 Medium

llama.cpp before b5649 contains a heap out-of-bounds read vulnerability in llamavocab::impl::printinfo() that allows attackers to crash…

CVE-2025-71439

llama.cpp before b5649 contains a heap out-of-bounds read vulnerability in llamavocab::impl::printinfo() that allows attackers to crash model loading via oversized special token IDs. Attackers can supply a crafted GGUF model with a short tokenizer.ggml.tokens array, causing llama-cli or llama-server to read past idtotoken and crash.

Part of the AI stack

This affects GGUF, llama.cpp, in Inference & serving, Runtimes & formats.

Assessment

CVSS: 6.8 (v4.0)
Severity: Medium
Known exploited: Not listed in CISA KEV
Exploit availability: Nothing found in Exploit-DB, Metasploit, Nuclei or public PoC repositories
Published: 27 October 2026
Last updated: 28 October 2026
Weakness: CWE-125
Assigner: disclosure@vulncheck.com

CVSS vectorCVSS:4.0/AV:L/AC:L/AT:N/PR:N/UI:P/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

At the source

References

Community proofs of concept

No community proof of concept yet. If you have one for a published vulnerability and the right to share it, add it below — it will appear here once reviewed.

Submit a proof of concept

Only for publicly disclosed vulnerabilities, and only material from testing you were authorised to perform. This platform does not buy, sell, or broker exploits — see earning from vulnerability research for legitimate bounty programs. Submissions are reviewed before they appear. Text only; do not include live credentials, personal data, or anything targeting systems you are not permitted to test.

Title *

Proof of concept *

Write-up (optional)

Affected versions (optional)

Name or handle (optional)

References (optional, one URL per line)

Licence for your contribution *

Private contact (optional, never shown)

Before you submit * This vulnerability is already publicly disclosed (it has a published CVE). This material comes only from testing I was authorised to perform. I have the legal right to publish this content and am not breaching any agreement by doing so. It contains no live credentials, personal data, or anything targeting systems I am not authorised to test. I understand it will be reviewed before it appears publicly and may be rejected. I agree to the submission terms.

What is 6 + 7? *

Query this programmatically

Agents can read this through the MCP endpoint, or over HTTP:

curl https://cve.zerocrew.in/api/cve/CVE-2025-71439