CVE-2026-100654
vLLM before 0.29.0 accepts user-controlled stop_token_ids on the OpenAI-compatible POST /v1/completions and POST /v1/chat/completions endpoints but validates o…
Auswirkung
vLLM before 0.29.0 accepts user-controlled stop_token_ids on the OpenAI-compatible POST /v1/completions and POST /v1/chat/completions endpoints but validates only that the values are integers, not that each token id is within the model vocabulary/logits range. When min_tokens > 0, the stop token ids are used as logits indices to suppress stop tokens, so an out-of-range id reaches a CUDA indexing operation (index_put_) and triggers a device-side assertion. An authenticated API user can send a single malformed completion request that returns 500 Internal Server Error and puts EngineCore into a fatal state, causing subsequent requests to fail until the service is restarted (denial of service).
Ein Angreifer kann die Lücke über das Netzwerk ausnutzen, mit einem normalen Benutzerkonto, ohne Zutun eines Nutzers. Bei Erfolg: hohes Risiko für Ausfälle.
Betroffene Produkte · Vllm
| Produkt | Betroffene Versionen |
|---|---|
| Vllm | vor 0.29.0 |