Vulnerability Snapshot
CVE-2026-61539 is rated CRITICAL — exploitation is trivial or already observed in the wild and impact is severe. Patch immediately, not on the next maintenance window.
Executive Summary
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.
Why It Matters
CVE-2026-61539 is rated CRITICAL severity, requiring immediate attention from security teams.
CVSS Base Score: 10/10
🛡️What Defenders Should Check
Use Vulnios to continuously monitor your exposure to CVE-2026-61539 and similar vulnerabilities.
References & Sources
How Vulnios Detects This
Vulnios scans for this vulnerability using Trivy and Grype for SBOM-based CVE matching and Vulnios CVE feed continuous monitoring against your asset inventory. Run a scan against your environment to see whether you are exposed; findings are linked back to the original CVE record so triage starts with the patch path already known.
AI Security Advisor
Powered by Gemini
Get AI-powered security recommendations tailored to this specific threat — including risk assessment, detection guidance, MITRE ATT&CK mapping, and actionable remediation steps.
Sources
Related Threat Alerts
Frequently Asked Questions
What is CVE-2026-61539?
CVE-2026-61539 is a critical-severity vulnerability tracked under the Common Vulnerabilities and Exposures program. 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
Am I affected?
Affected technology is listed in the Affected Products section above. If your asset inventory contains any of them, assume in-scope until you can prove otherwise.
How urgent is the response?
Critical: do not wait for your normal patch cycle. Verify exposure today, apply the vendor patch immediately, and add detection rules for any post-exploit indicators.
How do I remediate?
Apply the vendor patch listed in the upstream advisory linked under Sources. If the patch is not yet available, follow the vendor-supplied workaround (often a config flag or feature disable) and add detections for the published exploit pattern in your SIEM. Re-scan after the patch lands to confirm the finding clears.
Where can I track exploitation activity?
Watch CISA's Known Exploited Vulnerabilities catalog for CVE-2026-61539. Cross-reference with public exploit databases and your own SIEM/IDS for indicator-of-compromise patterns. Vulnios tracks KEV status automatically and surfaces it on the asset findings view.
How does Vulnios help with this?
Vulnios continuously cross-references your asset inventory against the live CVE feed (NVD, vendor advisories, CISA KEV, and curated OSINT). When a new CVE matches your environment, you get a prioritized finding with the severity, KEV status, exploit-prediction (EPSS), and a direct path to the vendor patch. You can start a free scan from the homepage.
Protect Your Organization
Monitor CVEs, scan for vulnerabilities, and get real-time threat alerts — all in one platform.
Get instant alerts on Telegram
Join our public channel for real-time critical CVE alerts.
Follow @vulnios