EUVD-2026-64146
GHSA-x2rj-828p-hx9m
Severity by source
AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
AV:N/PR:N because the endpoint is remote and unauthenticated by default; AC:H because exploitation depends on reliably steering the LLM to emit a malicious Python expression; S:U as code runs within the server process authority.
Primary rating from Vendor (GitHub_M).
CVSS VectorVendor: GitHub_M
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
Lifecycle Timeline
4DescriptionCVE.org
Xinference is an inference API for running open-source, speech, and multimodal models. In 2.5.0 and earlier, Xinference passes attacker-influenced Llama3 tool-call output to eval() in xinference/model/llm/tool_parsers/llama3_tool_parser.py and xinference/model/llm/utils.py. Requests to /v1/chat/completions with a tools field flow through xinference/api/restful_api.py, xinference/model/llm/transformers/core.py, handle_chat_result_non_streaming(), and _post_process_completion() before extract_tool_calls() or _eval_llama3_chat_arguments() evaluates the model-generated Python expression. An unauthenticated remote attacker can influence that output through a crafted prompt and execute commands in the Xinference server process context. This issue is fixed in version 2.7.0.
Articles & Coverage 2
AnalysisAI
Remote code execution in Xorbits Inference (Xinference) 2.5.0 and earlier lets a remote attacker run arbitrary OS commands in the server process by abusing an unsafe eval() in the Llama3 tool-call parser. Any request to the OpenAI-compatible /v1/chat/completions endpoint with a tools field triggers evaluation of model-generated Python, so an attacker who steers the model's output via a crafted prompt achieves full server compromise; because authentication is disabled in the default deployment, the vendor advisory rates this exploitable by an unauthenticated remote attacker. …
Unlock full vulnerability intelligence
- Risk assessment & exploitation conditions
- Attack chain visualization
- Remediation with exact patch versions
- Threat intelligence from 22 sources
- Personal watchlist & email alerts
Free forever · No credit card required
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata
Vulnerability AssessmentAI
| Exploitation | Requires the target to be using the Transformers backend serving a Llama3-style tool-calling model, and the attacker's request must include a tools field so the vulnerable tool-call post-processing path (_post_process_completion → extract_tool_calls/_eval_llama3_chat_arguments) is invoked. … Additional conditions and limiting factors are described in the full assessment. |
| Risk Assessment | Signals are strong but not uniform. … Full risk analysis with EPSS, KEV, and SSVC signal comparison available after sign-in. |
| Exploit Scenario | Full exploit scenario with step-by-step reproduction available after sign-in. |
| Remediation | Upgrade to Xinference 2.7.0, which replaces eval() with json.loads() plus an ast.literal_eval() fallback (safe literal parsing only) - Vendor-released patch: 2.7.0 (release https://github.com/xorbitsai/inference/releases/tag/v2.7.0; fix in PR https://github.com/xorbitsai/inference/pull/4786 and commit 1b3d220f342ce68d34cec4586d9409d457dadc42). … Detailed patch versions, workarounds, and compensating controls in full report. |
Recommended ActionAI
Within 24 hours, inventory all Xinference deployments, identify instances running versions 2.5.0 or earlier, and if immediate patching is not feasible, implement network access controls to restrict access to the /v1/chat/completions endpoint. …
Sign in for detailed remediation steps and compensating controls.
Threat intelligence, references, and detailed analysis are available after sign-in.
Wazuh SIEM platform versions 4.4.0 through 4.9.0 contain an unsafe deserialization vulnerability in the DistributedAPI t
BentoML version 1.4.2 and earlier contains an unauthenticated remote code execution vulnerability through insecure deser
pgAdmin 4 contains critical remote code execution vulnerabilities in the Query Tool download and Cloud Deployment endpoi
The renderLocalView function in render/views.py in graphite-web in Graphite 0.9.5 through 0.9.10 uses the pickle Python
BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Rated critica
OpenSSL before 0.9.8za, 1.0.0 before 1.0.0m, and 1.0.1 before 1.0.1h does not properly restrict processing of ChangeCiph
pyLoad download manager version prior to 0.5.0b3.dev77 exposes the Flask SECRET_KEY through an unauthenticated endpoint.
Langflow (a visual LLM pipeline builder) contains a critical unauthenticated code execution vulnerability (CVE-2026-3301
In Mercurial before 4.1.3, "hg serve --stdio" allows remote authenticated users to launch the Python debugger, and conse
Unauthenticated remote code execution in Marimo ≤0.20.4 allows attackers to execute arbitrary system commands via the `/
pyLoad is the free and open-source Download Manager written in pure Python. Rated medium severity (CVSS 5.3), this vulne
Cross-user flow execution in Langflow (< 1.9.1) lets any authenticated API-key holder run another user's flow by passing
Same weakness CWE-95 – Eval Injection
View allSame technique Information Disclosure
View allShare
External POC / Exploit Code
Leaving vuln.today