{"id":"GHSA-3cgq-c484-rx5j","kind":"vulnerability","aliases":["GHSA-3cgq-c484-rx5j"],"related":[],"title":"Duplicate Advisory: vLLM 0.7.3 before 0.31.0 DoS via Unvalidated mm_processor_kwargs","description":"## Duplicate Advisory\n\nThis advisory has been withdrawn because it is a duplicate of GHSA-gx7p-2j49-hfq4. This link is maintained to preserve external references.\n\n## Original Description\nvLLM from 0.7.3 before 0.31.0 contains an uncontrolled resource consumption vulnerability that allows remote clients to control multimodal processor parameters by supplying per-request mm_processor_kwargs. Attackers can set large image size, patch count, crop count, or frame count values in Isaac, InternVL, H2OVL, and other processors to exhaust memory and terminate the shared server.","summary":"vLLM, an LLM serving library, accepts per-request mm_processor_kwargs without validation, so a remote client can set huge image, patch, crop or frame counts. This exhausts memory and can crash the shared server for all users.","severity":"high","cvss":7.5,"epss":null,"exploited":false,"exploited_at":null,"products":[{"name":"vllm","fixed":[],"affected":">= 0.7.3, < 0.31.0","ecosystem":"pypi"}],"ecosystems":["pypi"],"tags":["ai-tooling","cwe-400","model-serving","python","self-hosted","vllm"],"not_affected_if":["multimodal-models-not-used","version-outside-range"],"how_to_check":"Run pip show vllm and check whether the version is >= 0.7.3 and < 0.31.0.","fix":"Upgrade vllm to 0.31.0 or later, per the duplicate advisory GHSA-gx7p-2j49-hfq4 which this entry points to.","status":"withdrawn","published_at":"2026-10-21T03:32:16+00:00","first_seen_at":"2026-10-23T20:59:05+00:00","triaged_at":"2026-10-23T21:00:17+00:00","updated_at":"2026-10-23T21:00:17+00:00","category":"ai-tooling","confidence":0.6,"triage":{"at":"2026-10-23T21:00:17Z","model":"claude-haiku-5-5","output":{"fix":"Upgrade vllm to 0.31.0 or later, per the duplicate advisory GHSA-gx7p-2j49-hfq4 which this entry points to.\nunknown","reach":"internet","impact":"medium","summary":"vLLM, an LLM serving library, accepts per-request mm_processor_kwargs without validation, so a remote client can set huge image, patch, crop or frame counts. This exhausts memory and can crash the shared server for all users.","category":"ai-tooling","who_cares":["vllm","python","self-hosted","model-serving"],"confidence":0.6,"needs_auth":"unknown","worst_case":"Remote clients crash the shared vLLM server by exhausting memory.","analyst_note":"Advisory is withdrawn as a duplicate; verify the fix and auth requirements in GHSA-gx7p-2j49-hfq4 before relying on this entry.","how_to_check":"Run pip show vllm and check whether the version is >= 0.7.3 and < 0.31.0.\nunknown","needs_analyst":true,"not_affected_if":["version-outside-range","multimodal-models-not-used"],"user_interaction":"no","needs_special_setup":"unknown","severity_if_unknown":"high"},"cost_usd":0.00051,"input_tokens":2961,"output_tokens":428,"severity_before":"high"},"analyst_status":null,"analyst_note":null,"search":"'-400':28B '0.31.0':6A,108C '0.7.3':4A,106C '2j49':92C '3cgq':15A,20A 'accept':42B 'advisori':2A,77C,79C 'ai':25B 'ai-tool':24B 'allow':116C 'attack':132C 'c484':16A,21A 'client':54B,118C 'consumpt':113C 'contain':109C 'control':120C 'count':63B,139C,141C,144C 'crash':69B 'crop':60B,140C 'cwe':27B 'descript':103C 'dos':7A 'duplic':1A,76C,87C 'exhaust':65B,154C 'extern':100C 'frame':62B,143C 'ghsa':14A,19A,90C 'ghsa-3cgq-c484-rx5j':13A,18A 'ghsa-gx7p-2j49-hfq4':89C 'gx7p':91C 'h2ovl':149C 'hfq4':93C 'host':35B 'huge':57B 'imag':58B,136C 'internvl':148C 'isaac':147C 'kwarg':12A,48B,131C 'larg':135C 'librari':41B 'link':95C 'llm':39B 'maintain':97C 'memori':66B,155C 'mm':10A,46B,129C 'model':30B 'model-serv':29B 'multimod':121C 'origin':102C 'paramet':123C 'patch':59B,138C 'per':44B,127C 'per-request':43B,126C 'preserv':99C 'processor':11A,47B,122C,130C,152C 'python':32B 'refer':101C 'remot':53B,117C 'request':45B,128C 'resourc':112C 'rx5j':17A,22A 'self':34B 'self-host':33B 'serv':31B,40B 'server':72B,160C 'set':56B,134C 'share':71B,159C 'size':137C 'suppli':125C 'termin':157C 'tool':26B 'uncontrol':111C 'unvalid':9A 'user':75B 'valid':50B 'valu':145C 'via':8A 'vllm':3A,23A,36B,37B,104C 'vulner':114C 'withdrawn':82C 'without':49B","nicknames":[],"mitigations":[],"who_affected":null,"tier":"B","verified":{},"analyst_revisit_at":null,"first_public_at":"2026-10-21T03:32:16+00:00","kev":null,"cvss_vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","cvss_by":"GitHub","poc_count":0,"sources":[{"source":"ghsa","external_id":"GHSA-3cgq-c484-rx5j","url":"https://github.com/advisories/GHSA-3cgq-c484-rx5j","severity":"high","published_at":"2026-10-21T03:32:16+00:00","modified_at":"2026-10-23T20:58:04+00:00","first_seen_at":"2026-10-23T20:59:05+00:00","fetched_at":"2026-10-23T20:59:05+00:00","note":null}],"documents":[],"notice":"Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai"}