llama.cpp heap out-of-bounds write in hrm_text model loader
Affects llama.cpp builds from b11003 up to but not including b12556 when loading a crafted hrm_text GGUF model. Fixed in b12556.
- Exploited
- No known exploitation
- Fix
- Fixed in b12556
- Who is exposed
- Specific setups only
Summary
llama.cpp from ggml-org has an integer overflow in its hrm_text model loader that lets a local attacker, without logging in, supply a crafted GGUF model file that writes past heap memory and could enable code execution. The attack requires a user to load the malicious model file.
Written by a model
Original description
llama.cpp b11003 before b12556 contains an integer overflow in the hrm_text loader load_arch_hparams() that allows attackers to bypass the block count check using crafted GGUF metadata. Attackers can supply a GGUF model whose layer factors wrap modulo 2^32, causing load_arch_tensors() to write tensor pointers past model.layers, corrupting heap memory and potentially enabling code execution.
Who should check
Teams running llama.cpp builds from b11003 up to b12556 that load HRM (hrm_text) GGUF models, including anyone accepting untrusted GGUF files
“Affecting llama.cpp >= b11003, < b12556”
What to do
Check whether your setup matches what the bug needs.
llama.cpp builds from b11003 up to but not including b12556 Loading an hrm_text (HRM) GGUF model whose layer factors are craftedUpgrade to a fixed version: b12556.
Check your systems with an agent
Paste this into a coding or ops agent. It checks your systems read-only and reports a table per asset.
Check whether my systems are affected by CVE-2026-133710 (llama.cpp b11003 before b12556 Heap Out-of-Bounds Write via hrm_text Model Loader), using the cve.now data below.
Rules: work read-only. Do not run exploit code, proof-of-concept scripts or scanners that send attack payloads, and do not change any configuration, package, image or deployment. Report findings only.
What cve.now has for CVE-2026-133710 (entry last updated 2026-11-20):
Affected:
- ggml-org llama.cpp: from b11003, fixed in b12556
Fixed versions: b12556.
Vendor patches: 50518b6 (source: https://www.vulncheck.com/advisories/llama-cpp-b11003-before-b12556-heap-out-of-bounds-write-via-hrm-text-model-loader). Check each asset for these patch ids, not only its version.
Conditions (all must be true for the bug to be hit):
- llama.cpp builds from b11003 up to but not including b12556 (source: https://www.vulncheck.com/advisories/llama-cpp-b11003-before-b12556-heap-out-of-bounds-write-via-hrm-text-model-loader)
- Loading an hrm_text (HRM) GGUF model whose layer factors are crafted (source: https://www.vulncheck.com/advisories/llama-cpp-b11003-before-b12556-heap-out-of-bounds-write-via-hrm-text-model-loader)
Setup: unknown whether the bug needs a non-default feature or setting. Check the advisory.
Attacker needs: local access, no login, a user action.
CISA KEV: CVE-2026-133710 is not listed as of cve.now's last read.
This KEV status is for CVE-2026-133710 only. If you also check another CVE, look up its own KEV status; never assume it from this one.
How to check:
- Installed software: check what is actually running, not what a repo declares. Read the version from each host, container image and Kubernetes node (the binary's version output, the OS package database, image layers), and check whether the conditions above hold there. Distribution packages may carry a backported fix under an older version number; confirm against the distribution's own advisory.
Verify against the linked advisories before concluding, and cite the date you read each one:
- https://github.com/ggml-org/llama.cpp/pull/32601
- https://github.com/ggml-org/llama.cpp/commit/50518b64c6f2191a116d2760042f106c4b3387d7
- https://github.com/ggml-org/llama.cpp/blob/b12555/src/models/hrm-text.cpp
- https://github.com/ggml-org/llama.cpp/releases/tag/b12556
- https://www.vulncheck.com/advisories/llama-cpp-b11003-before-b12556-heap-out-of-bounds-write-via-hrm-text-model-loader
Do not conclude an asset is unaffected unless you checked it and the evidence covers both its version and the conditions above. If you could not check something, mark it unknown and say what is missing.
Report a table with one row per asset and these columns: Asset | Version or build | Status (affected, fixed, not applicable, unknown) | Compensating controls | Exposure (internet-facing, internal, none) | Evidence | Next action.
Entry: https://cve.now/CVE-2026-133710
For follow-up questions, use the cve.now MCP server at https://mcp.cve.now/mcp (no authentication), for example execute with name cve.get and arguments {"id": "CVE-2026-133710"}.Affected versions
- ggml-org llama.cppFrom b11003 · fixed in b12556
cve.now Index
- Exploitation0/40No signal
- Exposure5/25Specific setups only
- Reach?/20Unknown. No tracked technology or package matched
- Severity9.5/10CVSS 8.5, total technical impact per CISA
- No patch0/5No signal
What would change the score
- Would score about 65 if added to CISA KEV
- Would score about 60 if CISA's SSVC rates exploitation active
- Would score about 25 if CISA's SSVC rates it automatable
Timeline
- CVE published
- First seen by cve.now
- CISA SSVC assessment
Exploitation: none · Automatable: no · Technical impact: total
- Record last updated
Evidence 7
| Kind | Source |
|---|---|
| CVE record 2Establishes: The identifier and description. | |
| Record | CVE List · CVE-2026-133710 |
| Record | NVD · CVE-2026-133710 |
| Other links 5 | |
| Patch | GitHub · ggml-org/llama.cpp · Pull request #32601 |
| Patch | GitHub · ggml-org/llama.cpp · Commit 50518b64 |
| Repository | GitHub · ggml-org/llama.cpp · src/models/hrm-text.cpp |
| Repository | GitHub · ggml-org/llama.cpp · Release b12556 |
| Link | vulncheck.com · advisories/llama-cpp-b11003-before-b12556-heap-out-of-bounds-write-via-hrm-text-model-loader |
Questions about CVE-2026-133710
Is CVE-2026-133710 being exploited?
It is not in CISA's Known Exploited Vulnerabilities catalog, and CISA's SSVC assessment does not mark exploitation as active. That does not mean it has not been attacked: vendors and researchers can report attacks before CISA lists them. No public exploit code is tracked.
Source: cve.now feed data
What does this mean?
Loading an hrm_text (HRM) GGUF model whose layer factors are crafted, and wrap modulo 2^32, corrupts heap memory through out-of-bounds writes. This can potentially enable code execution.
Source: vulncheck.com
Which versions are affected by CVE-2026-133710?
llama.cpp: from b11003 before b12556.
Source: cve.now feed data
Am I affected, and how do I check?
Builds of llama.cpp from b11003 up to but not including b12556 are in the affected range, and the bug is triggered by loading an hrm_text GGUF model with crafted layer factors. The build identifier of the llama.cpp binary in use shows which range it falls in.
Source: vulncheck.com
How do I fix CVE-2026-133710?
Upgrade to build b12556 or later, which contains the fix. The vendor's patch is commit 50518b6.
Source: vulncheck.com
Related
- CVE-2026-118375llama.cpp b8841 before b11865 Stack-based Buffer Overflow via ggml-rpc RDMA rdma_recv()
- CVE-2026-107183llama.cpp before b11393 Use-After-Free via common_chat_peg_mapper chat_parser
- CVE-2026-43632llama.cpp b7492–b9060 Use-After-Free in Tokenization Endpoints
- CVE-2026-43631llama.cpp b7492–b9060 Use-After-Free RCE via llama-server
- CVE-2026-43629llama.cpp b4882–b9058 Buffer Overflow in KV Cache State Restore
- CVE-2024-58400llama.cpp before b3701 Out-of-Bounds Read in llama_decode via Token IDs
No corrections so far. Something wrong on this page? Report an error.