Buffer overflow in turboderp ExLlamaV2 Safetensors tensor loader
Key facts
- Attack vector
- Network
- Privileges required
- None
- User interaction
- Passive
- Attack complexity
- Low
- Attack requirements
- None
- Impact type
- —
- Confidentiality · integrity · availability
- Low · Low · Low
- Fix status
- No fix listed
Attack profile from CVSS 4.0 · CNA. Fix status from CNA.
Summary
turboderp ExLlamaV2 versions 0.3.0, 0.3.1 and 0.3.2 are affected. The STFile.get_tensor function in the Safetensors tensor loader does not check the data_offsets argument, which lets a write go past the buffer (CWE-119). An unauthenticated remote attacker can trigger it with some user interaction. A public exploit is available.
Written with Claude Haiku 5.5 from the sources below · 2026-10-28
Description from the CNA
A vulnerability was detected in turboderp ExLlamaV2 up to 0.3.2. Affected is the function STFile.get_tensor of the file exllamav2/stloader.py of the component Safetensors Tensor Loader. Performing a manipulation of the argument data_offsets results in buffer overflow. The attack is possible to be carried out remotely. The exploit is now public and may be used. The project was informed of the problem early through an issue report but has not responded yet.
Why it matters now
SSVC decision points
CISA ADP · SSVC 2.0.3
- Exploitation
- Poc
- Automatable
- Yes
- Technical impact
- Partial
The decision (Track, Attend, Act) also depends on your mission and well-being impact; see CISA's SSVC guide.
Affected products
Remediation
The sources list no fixed version for ExLlamaV2.
Written with Claude Haiku 5.5 from the sources below · 2026-10-28
Timeline
- 2026-10-15 Threatist record created from CVE-2026-110695
- 2026-10-15 CISA SSVC decision points changed
- 2026-10-15 Priority changed: LOW → MEDIUM
References
Sources
This product uses the NVD API but is not endorsed or certified by the NVD. Sources and licences