CVE-2026-100653
Description
vLLM is an inference and serving engine for large language models. In versions from 0.22.1 through 0.28.0, the operator-supplied model revision pin (--revision / --code-revision) is not propagated to several Hugging Face artifact loads for the FunAudioChat and Tarsier2 architectures: the WhisperFeatureExtractor and speech_tokenizer PreTrainedTokenizerFast loads in vllm/model_executor/models/funaudiochat.py and the Qwen2VLConfig.from_pretrained call used by Tarsier2ProcessingInfo in vllm/model_executor/models/qwen2_vl.py. As a result, deployments pinned to a reviewed revision still resolve these behavior-affecting processor, tokenizer, and config artifacts from the repository's default revision, so a later change to the upstream default branch can alter audio preprocessing, speech tokenizer behavior, or Tarsier2 configuration without any change to the operator's configured pin. This is a supply-chain integrity and reproducibility failure for pinned deployments; it is residual to the earlier fix tracked as GHSA-3ww4-5jv9-j5gm / CVE-2026-47155 and does not constitute remote code execution or a trust_remote_code=False bypass. The issue is fixed in version 0.28.0.
Weakness
- CWE-348: Use of Less Trusted Source
CISA SSVC decision points
| Exploitation | poc |
|---|---|
| Automatable | no |
| Technical impact | partial |
| Scored | 2026-09-30 |
Timeline
- 2026-09-26Published by the CNA13:23 UTC
- 2026-09-30CISA SSVC decision recorded4 days after publication
Affected products
Products listed as vulnerable. Upgrade to the version under "Fixed in", or a later one.
| Product | Vulnerable versions | Fixed in |
|---|---|---|
| vllm vllm | from 0.22.1 before 0.28.0fixed in 0.28.0 | 0.28.0 |
References
- https://github.com/vllm-project/vllm/commit/d26a28ab033697f55a1414b5b0435de7cd6045b6
- https://github.com/vllm-project/vllm/security/advisories/GHSA-hhv2-872h-628q
- https://www.vulncheck.com/advisories/vllm-0.22.1-before-0.28.0-incomplete-artifact-pin-propagation
Get more CVE data from the API.
curl 'https://api.contrastcyber.com/cve_lookup/CVE-2026-100653'curl 'https://api.contrastcyber.com/cve_risk_score/CVE-2026-100653'curl 'https://api.contrastcyber.com/cve_exploit/CVE-2026-100653'curl 'https://api.contrastcyber.com/cve_products/CVE-2026-100653'
Sources
- Vulnerability data from the National Vulnerability Database (NIST) and the CVE Program. CVE records: Copyright © 1999-2026, The MITRE Corporation, used under the CVE Program Terms of Use. CVE® is a registered trademark of The MITRE Corporation.
- CWE definitions: Copyright © 2006-2026, The MITRE Corporation, used under the CWE Terms of Use. CWE is a trademark of The MITRE Corporation.
- OWASP Top 10 categories by the OWASP Foundation (CC BY 3.0).
- EPSS scores by FIRST.org EPSS, used with attribution.
- KEV catalog and SSVC decisions by CISA.
- Advisory data from the GitHub Advisory Database through OSV.dev (CC BY 4.0).
- Exploit metadata from the Exploit Database (GPL-2.0-only, metadata only).
- Nuclei template names from nuclei-templates (MIT).