https://cve.radiocsirt.org Contains only the most 10 recent entries. http://www.rssboard.org/rss-specification python-feedgen en Fri, 02 Oct 2026 19:51:38 +0000 https://cve.radiocsirt.org/vuln/euvd-2026-329179 EUVD-2026-329179 https://cve.radiocsirt.org/vuln/euvd-2026-329179 https://cve.radiocsirt.org/vuln/fkie_cve-2026-54233 <p>vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.23.1rc0, vLLM&#39;s /v1/audio/transcriptions endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to ~14.9GB of float32 PCM at decode time. This vulnerability is fixed in 0.23.1rc0.</p> <p>vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.23.1rc0, vLLM&#39;s /v1/audio/transcriptions endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to ~14.9GB of float32 PCM at decode time. This vulnerability is fixed in 0.23.1rc0.</p> https://cve.radiocsirt.org/vuln/fkie_cve-2026-54233 https://cve.radiocsirt.org/vuln/ghsa-6pr9-rp53-2pmc <p><strong>Affected:</strong> PyPI: vllm</p> <p>### Summary vLLM&#39;s `/v1/audio/transcriptions` endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to ~14.9GB of float32 PCM at decode time. Tested on vLLM v0.19.0.</p> <p>### Details `SpeechToTextProcessor` rejects uploads over `VLLM_MAX_AUDIO_CLIP_FILESIZE_MB` (default 25MB) based on compressed byte length, but the audio decoder in `audio.py` accumulates all decoded frames into memory with no size limit before returning:</p> <p>```python # speech_to_text.py L184-189 if len(audio_data) / 1024 ** 2 &gt; self.max_audio_filesize_mb: raise VLLMValidationError(...) y, sr = load_audio(buf, sr=self.asr_config.sample_rate) # decoded size unchecked</p> <p># audio.py L77-107 chunks: list[npt.NDArray] = [] for frame in container.decode(stream): chunks.append(frame.to_ndarray()) audio = np.concatenate(chunks, axis=-1).astype(np.float32) # single contiguous allocation ```</p> <p>A 25MB OPUS file at 6kbps encodes ~8.7 hours of audio. Decoding produces ~5.7GB of float32 PCM (232x amplification), and `np.concatenate` then allocates a second contiguous array, bringing peak RSS to ~14.9GB from a single request. `SpeechToTextConfig.max_audio_clip_s` (default 30s) applies only after the full decode and does not prevent the allocation.</p> <p>### Impact An unauthenticated attacker can exhaust server memory with a small number of concurrent requests, each a valid upload within the documented size limit. Severity was assessed with reference to prior OOM vulnerability reports in vL…</p> <p><strong>Affected:</strong> PyPI: vllm</p> <p>### Summary vLLM&#39;s `/v1/audio/transcriptions` endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to ~14.9GB of float32 PCM at decode time. Tested on vLLM v0.19.0.</p> <p>### Details `SpeechToTextProcessor` rejects uploads over `VLLM_MAX_AUDIO_CLIP_FILESIZE_MB` (default 25MB) based on compressed byte length, but the audio decoder in `audio.py` accumulates all decoded frames into memory with no size limit before returning:</p> <p>```python # speech_to_text.py L184-189 if len(audio_data) / 1024 ** 2 &gt; self.max_audio_filesize_mb: raise VLLMValidationError(...) y, sr = load_audio(buf, sr=self.asr_config.sample_rate) # decoded size unchecked</p> <p># audio.py L77-107 chunks: list[npt.NDArray] = [] for frame in container.decode(stream): chunks.append(frame.to_ndarray()) audio = np.concatenate(chunks, axis=-1).astype(np.float32) # single contiguous allocation ```</p> <p>A 25MB OPUS file at 6kbps encodes ~8.7 hours of audio. Decoding produces ~5.7GB of float32 PCM (232x amplification), and `np.concatenate` then allocates a second contiguous array, bringing peak RSS to ~14.9GB from a single request. `SpeechToTextConfig.max_audio_clip_s` (default 30s) applies only after the full decode and does not prevent the allocation.</p> <p>### Impact An unauthenticated attacker can exhaust server memory with a small number of concurrent requests, each a valid upload within the documented size limit. Severity was assessed with reference to prior OOM vulnerability reports in vL…</p> https://cve.radiocsirt.org/vuln/ghsa-6pr9-rp53-2pmc https://cve.radiocsirt.org/vuln/pysec-2026-3404 <p><strong>Affected:</strong> PyPI: vllm</p> <p>### Summary vLLM&#39;s `/v1/audio/transcriptions` endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to ~14.9GB of float32 PCM at decode time. Tested on vLLM v0.19.0.</p> <p>### Details `SpeechToTextProcessor` rejects uploads over `VLLM_MAX_AUDIO_CLIP_FILESIZE_MB` (default 25MB) based on compressed byte length, but the audio decoder in `audio.py` accumulates all decoded frames into memory with no size limit before returning:</p> <p>```python # speech_to_text.py L184-189 if len(audio_data) / 1024 ** 2 &gt; self.max_audio_filesize_mb: raise VLLMValidationError(...) y, sr = load_audio(buf, sr=self.asr_config.sample_rate) # decoded size unchecked</p> <p># audio.py L77-107 chunks: list[npt.NDArray] = [] for frame in container.decode(stream): chunks.append(frame.to_ndarray()) audio = np.concatenate(chunks, axis=-1).astype(np.float32) # single contiguous allocation ```</p> <p>A 25MB OPUS file at 6kbps encodes ~8.7 hours of audio. Decoding produces ~5.7GB of float32 PCM (232x amplification), and `np.concatenate` then allocates a second contiguous array, bringing peak RSS to ~14.9GB from a single request. `SpeechToTextConfig.max_audio_clip_s` (default 30s) applies only after the full decode and does not prevent the allocation.</p> <p>### Impact An unauthenticated attacker can exhaust server memory with a small number of concurrent requests, each a valid upload within the documented size limit. Severity was assessed with reference to prior OOM vulnerability reports in vL…</p> <p><strong>Affected:</strong> PyPI: vllm</p> <p>### Summary vLLM&#39;s `/v1/audio/transcriptions` endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to ~14.9GB of float32 PCM at decode time. Tested on vLLM v0.19.0.</p> <p>### Details `SpeechToTextProcessor` rejects uploads over `VLLM_MAX_AUDIO_CLIP_FILESIZE_MB` (default 25MB) based on compressed byte length, but the audio decoder in `audio.py` accumulates all decoded frames into memory with no size limit before returning:</p> <p>```python # speech_to_text.py L184-189 if len(audio_data) / 1024 ** 2 &gt; self.max_audio_filesize_mb: raise VLLMValidationError(...) y, sr = load_audio(buf, sr=self.asr_config.sample_rate) # decoded size unchecked</p> <p># audio.py L77-107 chunks: list[npt.NDArray] = [] for frame in container.decode(stream): chunks.append(frame.to_ndarray()) audio = np.concatenate(chunks, axis=-1).astype(np.float32) # single contiguous allocation ```</p> <p>A 25MB OPUS file at 6kbps encodes ~8.7 hours of audio. Decoding produces ~5.7GB of float32 PCM (232x amplification), and `np.concatenate` then allocates a second contiguous array, bringing peak RSS to ~14.9GB from a single request. `SpeechToTextConfig.max_audio_clip_s` (default 30s) applies only after the full decode and does not prevent the allocation.</p> <p>### Impact An unauthenticated attacker can exhaust server memory with a small number of concurrent requests, each a valid upload within the documented size limit. Severity was assessed with reference to prior OOM vulnerability reports in vL…</p> https://cve.radiocsirt.org/vuln/pysec-2026-3404 https://cve.radiocsirt.org/vuln/wid-sec-w-2026-1897 <p>Ein Angreifer kann mehrere Schwachstellen in vllm ausnutzen, um Sicherheitsmaßnahmen zu umgehen, einen Denial-of-Service-Zustand zu verursachen, Daten zu manipulieren oder vertrauliche Informationen offenzulegen.</p> <p>Ein Angreifer kann mehrere Schwachstellen in vllm ausnutzen, um Sicherheitsmaßnahmen zu umgehen, einen Denial-of-Service-Zustand zu verursachen, Daten zu manipulieren oder vertrauliche Informationen offenzulegen.</p> https://cve.radiocsirt.org/vuln/wid-sec-w-2026-1897