{"id":"CVE-2026-116173","kind":"vulnerability","aliases":["CVE-2026-116173"],"related":[],"title":"vLLM: oversized logprob requests over gRPC crash the inference engine","description":"vLLM from 0.22.0 before 0.24.0 contains a denial of service vulnerability that allows remote attackers to terminate EngineCore by bypassing logprob_token_ids limits in the Rust gRPC frontend. Attackers can send Generate or GenerateStream requests with output_logprobs enabled and more than 128 output_candidates.token_ids, triggering a fatal sampler exception that crashes the engine.","summary":"vLLM from 0.22.0 before 0.24.0 has a denial of service flaw in its Rust gRPC frontend. A remote request with output logprobs enabled and more than 128 candidate token IDs makes the sampler throw a fatal exception and terminates EngineCore.","severity":"high","cvss":8.2,"epss":null,"exploited":false,"exploited_at":null,"products":[],"ecosystems":[],"tags":["ai-tooling","grpc","inference","python","self-hosted","vllm"],"not_affected_if":["default-config","internal-only"],"how_to_check":"- Check the installed vLLM version with `pip show vllm` and see whether it falls in 0.22.0 up to but not including 0.24.0.\n- Check whether the gRPC frontend is exposed: look for the Generate and GenerateStream endpoints reachable from outside the host.","fix":"- Upgrade vLLM to `0.24.0` or later.\n- Until upgraded, block external access to the gRPC Generate and GenerateStream endpoints.","status":"open","published_at":"2026-10-25T14:16:56+00:00","first_seen_at":"2026-10-25T14:41:23+00:00","triaged_at":"2026-10-25T14:42:26+00:00","updated_at":"2026-10-25T14:42:26+00:00","category":"ai-tooling","confidence":0.6,"triage":{"at":"2026-10-25T14:42:26Z","model":"claude-haiku-5-5","output":{"fix":"- Upgrade vLLM to `0.24.0` or later.\n- Until upgraded, block external access to the gRPC Generate and GenerateStream endpoints.","reach":"network","title":"vLLM: oversized logprob requests over gRPC crash the inference engine","impact":"low","summary":"vLLM from 0.22.0 before 0.24.0 has a denial of service flaw in its Rust gRPC frontend. A remote request with output logprobs enabled and more than 128 candidate token IDs makes the sampler throw a fatal exception and terminates EngineCore.","category":"ai-tooling","nicknames":[],"who_cares":["vllm","python","self-hosted","inference","grpc"],"confidence":0.6,"needs_auth":"unknown","worst_case":"Remote requests crash the inference engine, causing service outage.","analyst_note":"Confirm from the NVD entry or the vLLM advisory whether the gRPC frontend is enabled by default and whether authentication is required.","how_to_check":"- Check the installed vLLM version with `pip show vllm` and see whether it falls in 0.22.0 up to but not including 0.24.0.\n- Check whether the gRPC frontend is exposed: look for the Generate and GenerateStream endpoints reachable from outside the host.","needs_analyst":true,"not_affected_if":["default-config","internal-only"],"who_is_affected":"Teams running vLLM 0.22.x or 0.23.x with the Rust gRPC frontend reachable by clients they do not fully trust. Internal-only deployments with no untrusted callers are at lower risk. Teams on 0.24.0 or later, or on other vLLM versions outside the range, can skip it.","user_interaction":"no","needs_special_setup":"unknown","severity_if_unknown":"high"},"cost_usd":0.000648,"input_tokens":3366,"output_tokens":622,"severity_before":"high"},"analyst_status":null,"analyst_note":null,"search":"'-116173':13A,16A '-2026':12A,15A '0.22.0':29B,69C '0.24.0':31B,71C '128':53B,110C 'ai':18B 'ai-tool':17B 'allow':79C 'attack':81C,96C 'bypass':86C 'candid':54B 'contain':72C 'crash':7A,119C 'cve':11A,14A 'denial':34B,74C 'enabl':49B,106C 'engin':10A,121C 'enginecor':66B,84C 'except':63B,117C 'fatal':62B,115C 'flaw':37B 'frontend':42B,95C 'generat':99C 'generatestream':101C 'grpc':6A,20B,41B,94C 'host':25B 'id':56B,89C,112C 'infer':9A,21B 'limit':90C 'logprob':3A,48B,87C,105C 'make':57B 'output':47B,104C 'output_candidates.token':111C 'overs':2A 'python':22B 'remot':44B,80C 'request':4A,45B,102C 'rust':40B,93C 'sampler':59B,116C 'self':24B 'self-host':23B 'send':98C 'servic':36B,76C 'termin':65B,83C 'throw':60B 'token':55B,88C 'tool':19B 'trigger':113C 'vllm':1A,26B,27B,67C 'vulner':77C","nicknames":[],"mitigations":[],"who_affected":"Teams running vLLM 0.22.x or 0.23.x with the Rust gRPC frontend reachable by clients they do not fully trust. Internal-only deployments with no untrusted callers are at lower risk. Teams on 0.24.0 or later, or on other vLLM versions outside the range, can skip it.","tier":"B","verified":{},"analyst_revisit_at":null,"first_public_at":"2026-10-25T14:16:56+00:00","kev":null,"cvss_vector":"CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","cvss_by":"disclosure@vulncheck.com","poc_count":0,"sources":[{"source":"nvd","external_id":"CVE-2026-116173","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-116173","severity":null,"published_at":"2026-10-25T14:16:56+00:00","modified_at":"2026-10-25T14:16:56+00:00","first_seen_at":"2026-10-25T14:41:23+00:00","fetched_at":"2026-10-25T14:41:23+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"}