CVE-2026-103042

waiting for further signalsZero needs more public information before it can validate it.

GHSA-mqxw-j6vh-r94f ↗EUVD-2026-89483 ↗

no check availablenot proven exploitable
cve.orgpublic pocExploit code has been published by someone. Source signals can also carry a PoC (Proof of Concept).kevAttacks confirmed in the real world.
ModelTC LightLLM <= 1.2.0
CWE-770 Allocation of Resources Without Limits or Throttling
ETK-2026-36, permanent identifier
5d 09h 38msince last run
13mAI analysisAnalysis: how long the AI spent reversing this signal.
timeline
0 = trigger signal
trigger signal (cve.org) +0hwaiting for further signals+379h45m
0
+2h
+4h
+6h
+8h
+10h

description

LightLLM through version 1.2.0 is affected when it is launched with NCCL PD-transfer mode enabled. An unauthenticated remote party may consume unbounded memory in KV-transfer workers, causing worker crashes and potentially disrupting the node.

mitigation

Avoid enabling --pd_trans_mode nccl where unauthenticated network access is possible. Restrict access to the NCCL control channel to trusted hosts until an upstream fixed version is available.

check
no
exploit
no
automationDeterministic: a check that runs on its own, the same request on every target. Contextual: an agentic technique that needs to be applied per target, for bugs that depend on contextual flows.
contextual
auth requiredWhether an attacker needs an account first, based on the CVSS vector's Privileges Required.
no
EPSSEstimate of the chance this CVE is exploited in the next 30 days. Refreshed daily.
0.52%43rd percentile
agent run time
13m
cve.org record, t = 0The moment cve.org publishes the CVE in full.
2026-09-29 22:51Z
Δ to public pocExploit code has been published by someone. Source signals can also carry a PoC (Proof of Concept).
–
Δ to kevAttacks confirmed in the real world.
–

outcome

Contextual: an exploit technique exists, but no deterministic check could cover this one.

source signals

Public signals that triggered or updated the pipeline. These are inputs for the deltas above.

  1. 2026-09-29 22:51Z
    cve_org: trigger signal, started the run

this entry in the feed

GET /api/v1/feed/CVE-2026-103042

{
  "etk_id": "ETK-2026-36",
  "cve_id": "CVE-2026-103042",
  "display_id": "CVE-2026-103042",
  "ghsa_id": "GHSA-mqxw-j6vh-r94f",
  "euvd_id": "EUVD-2026-89483",
  "status": "published",
  "anchor_state": "anchored",
  "title": "NCCL control-channel memory exhaustion in LightLLM",
  "summary": "LightLLM through version 1.2.0 is affected when it is launched with NCCL PD-transfer mode enabled. An unauthenticated remote party may consume unbounded memory in KV-transfer workers, causing worker crashes and potentially disrupting the node.",
  "mitigation": "Avoid enabling `--pd_trans_mode nccl` where unauthenticated network access is possible. Restrict access to the NCCL control channel to trusted hosts until an upstream fixed version is available.",
  "vendor": "ModelTC",
  "product": "LightLLM",
  "affected": [
    {
      "affected_notes": [],
      "affected_ranges": [
        ">= 0, <= 1.2.0"
      ],
      "ecosystem": null,
      "fixed_versions": [],
      "package_name": "LightLLM",
      "purl": null
    }
  ],
  "cwes": [
    770
  ],
  "cvss_score": 8.7,
  "cvss_src": "cna_vendor",
  "cvss_vector": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N",
  "epss_score": 0.0052,
  "epss_percentile": 0.4302,
  "epss_asof": "2026-10-20",
  "automation": "contextual",
  "check": false,
  "check_confidence_basis": null,
  "exploit": false,
  "pipeline_minutes": 12.64217885,
  "t0_at": "2026-09-29T22:51:01.631000+00:00",
  "signals": {
    "trigger_signal": {
      "at": "2026-09-29T22:51:01.631000+00:00",
      "days": 0,
      "is_lower_bound": false,
      "measured_at": "2026-09-29T22:51:01.631000+00:00",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-103042"
    }
  }
}

published 2026-09-29 23:28Z
last change 2026-10-21 03:59Z

Need the technical details behind an entry?

Every entry here shows what we attempted and when. The technical details, the generated exploit that fired it and the evidence behind the verdict are not public. Tell us about your use case and we’ll get back to you about extended API access.