# CVE-2026-93839: LightLLM authentication bypass in /pd_register WebSocket endpoint

Source: https://junglewise.ai/threats/cve-2026-93839-lightllm-through-1-2-0-contains-an-authentication-bypass  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-09-18. Updated: 2026-09-26  
License: CC BY 4.0 with a link back

CVE-2026-93839 · Severity: critical · CVSS 9.8 · Published 2026-09-18

LightLLM is a Python-based framework for serving large language models. An unauthenticated attacker can register a rogue compute node via the /pd_register WebSocket endpoint, allowing them to intercept all user prompts (including sensitive data like API keys and personal information), cause complete service outages, or use the service to make requests to internal networks. The attack requires only network access to the endpoint and no valid credentials.

Technologies: [ModelTC LightLLM](https://junglewise.ai/threats/technologies/lightllm). Vendors: [ModelTC](https://junglewise.ai/threats/vendors/modeltc).

## Quick answers

**Is CVE-2026-93839 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-93839 being exploited in the wild.

**How severe is CVE-2026-93839?** Critical severity, with a CVSS score of 9.8, published 18 September 2026.

**Which products and versions are affected?** ModelTC LightLLM through 1.2.0.

**Is there a fix for CVE-2026-93839?** No fixed version is recorded yet. Check the vendor's advisory in the references below. If you run it, Junglewise drafts a mitigation that fits your setup and applies it once you approve, with senior security engineers behind it.

**How is CVE-2026-93839 attacked?** It is attacked over the network, in a software library.

**Am I affected by CVE-2026-93839?** Check whether you run ModelTC LightLLM at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

The /pd_register WebSocket endpoint accepts connections without authentication and accepts node registration JSON without validating that the declared client address belongs to the attacker. The vulnerability permits silent replacement of legitimate nodes in the routing pool via list-filter logic, enabling prompt interception and request hijacking. Additionally, the unvalidated client_ip_port is interpolated into health-check URLs, enabling SSRF attacks against internal addresses.

## Affected products

- ModelTC LightLLM through 1.2.0

## Timeline

- 2026-09-18: disclosed

## References

- [https://github.com/ModelTC/LightLLM](https://github.com/ModelTC/LightLLM)
- [https://github.com/ModelTC/LightLLM/issues/1576](https://github.com/ModelTC/LightLLM/issues/1576)

## Related threats

- [LightLLM denial of service via pd_kv_trans_params unpickling](https://junglewise.ai/threats/cve-2026-111160-lightllm-through-1-2-0-contains-a-denial-of-service)
- [LightLLM visual_only unauthenticated RCE via RPyC pickle deserialization](https://junglewise.ai/threats/cve-2026-103395-lightllm-through-1-2-0-visual-only-deployments-expose-an)
- [LightLLM authentication bypass in reinforcement learning control API](https://junglewise.ai/threats/cve-2026-103270-lightllm-through-1-2-0-mounts-reinforcement-learning-control)
- [LightLLM server-side request forgery in multimodal endpoints](https://junglewise.ai/threats/cve-2026-103243-lightllm-through-1-2-0-fails-to-validate-image-url-and-audio-url)
- [LightLLM memory exhaustion in NCCL control channel](https://junglewise.ai/threats/cve-2026-103042-lightllm-through-1-2-0-contains-a-memory-exhaustion)