| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| vLLM is an inference and serving engine for large language models (LLMs). From 0.16.0 to before 0.19.0, a server-side request forgery (SSRF) vulnerability in download_bytes_from_url allows any actor who can control batch input JSON to make the vLLM batch runner issue arbitrary HTTP/HTTPS requests from the server, without any URL validation or domain restrictions.
This can be used to target internal services (e.g. cloud metadata endpoints or internal HTTP APIs) reachable from the vLLM host. This vulnerability is fixed in 0.19.0. |
| vLLM is an inference and serving engine for large language models (LLMs). From 0.7.0 to before 0.19.0, the VideoMediaIO.load_base64() method at vllm/multimodal/media/video.py splits video/jpeg data URLs by comma to extract individual JPEG frames, but does not enforce a frame count limit. The num_frames parameter (default: 32), which is enforced by the load_bytes() code path, is completely bypassed in the video/jpeg base64 path. An attacker can send a single API request containing thousands of comma-separated base64-encoded JPEG frames, causing the server to decode all frames into memory and crash with OOM. This vulnerability is fixed in 0.19.0. |
| vLLM is an inference and serving engine for large language models (LLMs). From 0.1.0 to before 0.19.0, a Denial of Service vulnerability exists in the vLLM OpenAI-compatible API server. Due to the lack of an upper bound validation on the n parameter in the ChatCompletionRequest and CompletionRequest Pydantic models, an unauthenticated attacker can send a single HTTP request with an astronomically large n value. This completely blocks the Python asyncio event loop and causes immediate Out-Of-Memory crashes by allocating millions of request object copies in the heap before the request even reaches the scheduling queue. This vulnerability is fixed in 0.19.0. |
| A bug in WebAssembly code generation could have lead to a crash. It may have been possible for an attacker to leverage this to achieve code execution. This vulnerability was fixed in Firefox 135, Firefox ESR 128.7, Thunderbird 128.7, and Thunderbird 135. |
| The z-order of the browser windows could be manipulated to hide the fullscreen notification. This could potentially be leveraged to perform a spoofing attack. This vulnerability was fixed in Firefox 135 and Thunderbird 135. |
| A race condition could have led to private browsing tabs being opened in normal browsing windows. This could have resulted in a potential privacy leak. This vulnerability was fixed in Firefox 135, Firefox ESR 128.7, Thunderbird 128.7, and Thunderbird 135. |
| Certificate length was not properly checked when added to a certificate store. In practice only trusted data was processed. This vulnerability was fixed in Firefox 135, Firefox ESR 128.7, Thunderbird 128.7, and Thunderbird 135. |
| Memory safety bugs present in Firefox 134, Thunderbird 134, Firefox ESR 115.19, Firefox ESR 128.6, Thunderbird 115.19, and Thunderbird 128.6. Some of these bugs showed evidence of memory corruption and we presume that with enough effort some of these could have been exploited to run arbitrary code. This vulnerability was fixed in Firefox 135, Firefox ESR 115.20, Firefox ESR 128.7, Thunderbird 128.7, and Thunderbird 135. |
| The Thunderbird Address Book URI fields contained unsanitized links. This could be used by an attacker to create and export an address book containing a malicious payload in a field. For example, in the “Other” field of the Instant Messaging section. If another user imported the address book, clicking on the link could result in opening a web page inside Thunderbird, and that page could execute (unprivileged) JavaScript. This vulnerability was fixed in Thunderbird 128.7 and Thunderbird 135. |
| Memory safety bugs present in Firefox 135. Some of these bugs showed evidence of memory corruption and we presume that with enough effort some of these could have been exploited to run arbitrary code. This vulnerability was fixed in Firefox 135.0.1. |
| Android apps can load web pages using the Custom Tabs feature. This feature supports a transition animation that could have been used to trick a user into granting sensitive permissions by hiding what the user was actually clicking. This vulnerability was fixed in Firefox 136. |
| A select option could partially obscure the confirmation prompt shown before launching external apps. This could be used to trick a user in to launching an external app unexpectedly.
*This issue only affects Android versions of Firefox.*. This vulnerability was fixed in Firefox 136. |
| Under certain circumstances, a user opt-in setting that Focus should require authentication before use could have been be bypassed (distinct from CVE-2025-0245). This vulnerability was fixed in Firefox 136. |
| When String.toUpperCase() caused a string to get longer it was possible for uninitialized memory to be incorporated into the result string. This vulnerability was fixed in Firefox 136 and Thunderbird 136. |
| jar: URLs retrieve local file content packaged in a ZIP archive. The null and everything after it was ignored when retrieving the content from the archive, but the fake extension after the null was used to determine the type of content. This could have been used to hide code in a web extension disguised as something else like an image. This vulnerability was fixed in Firefox 136, Firefox ESR 128.8, Thunderbird 136, and Thunderbird 128.8. |
| Memory safety bugs present in Firefox 135, Thunderbird 135, Firefox ESR 128.7, and Thunderbird 128.7. Some of these bugs showed evidence of memory corruption and we presume that with enough effort some of these could have been exploited to run arbitrary code. This vulnerability was fixed in Firefox 136, Firefox ESR 128.8, Thunderbird 136, and Thunderbird 128.8. |
| Memory safety bugs present in Firefox 135 and Thunderbird 135. Some of these bugs showed evidence of memory corruption and we presume that with enough effort some of these could have been exploited to run arbitrary code. This vulnerability was fixed in Firefox 136 and Thunderbird 136. |
| Malicious websites utilizing a server-side redirect to an internal error page could result in a spoofed website URL. This vulnerability was fixed in Firefox for iOS 136. |
| Websites redirecting to a non-HTTP scheme URL could allow a website address to be spoofed for a malicious page. This vulnerability was fixed in Firefox for iOS 136. |
| When requesting an OpenPGP key from a WKD server, an incorrect padding size was used and a network observer could have learned the length of the requested email address. This vulnerability was fixed in Thunderbird 136 and Thunderbird 128.8. |
