### Summary Users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct `ndim` but incorrect `shape` (e.g. hidden dimension is wrong), regardless of whether the model is intended to support such inputs (as defined in the Supported Models page). The issue has existed ever since we added support for image embedding inputs, i.e. #6613 (released in v0.5.5) ### Details Using image embeddings as an example: - For models that support image embedding inputs, the engine crashes when scattering the embeddings to `inputs_embeds` (mismatched shape) - For models that don't support image embedding inputs, the engine crashes when validating the inputs inside `get_input_embeddings` (validation fails). This happens because we only validate `ndim` of the tensor, but not the full shape, in input processor (via `MultiModalDataParser`). ### Impact - Denial of service by crashing the engine ### Mitigation - Use API key to limit access to trusted users. - Set `--limit-mm-per-prompt` to 0 for all non-text modalities to ban multimodal inputs, which includes multimodal embedding inputs. However, the model would then only accept text, defeating the purpose of using a multi-modal model. ### Resolution - https://github.com/vllm-project/vllm/pull/27204
### Summary Users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct `ndim` but incorrect `shape` (e.g. hidden dimension is wrong), regardless of whether the model is intended to support such inputs (as defined in the Supported Models page). The issue has existed ever since we added support for image embedding inputs, i.e. #6613 (released in v0.5.5) ### Details Using image embeddings as an example: - For models that support image embedding inputs, the engine crashes when scattering the embeddings to `inputs_embeds` (mismatched shape) - For models that don't support image embedding inputs, the engine crashes when validating the inputs inside `get_input_embeddings` (validation fails). This happens because we only validate `ndim` of the tensor, but not the full shape, in input processor (via `MultiModalDataParser`). ### Impact - Denial of service by crashing the engine ### Mitigation - Use API key to limit access to trusted users. - Set `--limit-mm-per-prompt` to 0 for all non-text modalities to ban multimodal inputs, which includes multimodal embedding inputs. However, the model would then only accept text, defeating the purpose of using a multi-modal model. ### Resolution - https://github.com/vllm-project/vllm/pull/27204
Update vllm to 0.11.1 if you use the affected versions. Test the change in a non-production environment first.
Local check
hol-guard supply-chain scanvLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs affects vllm (pip). Severity is high. ### Summary Users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct `ndim` but incorrect `shape` (e.g. hidden dimension is wrong), regardless of whether the model is intended to support such inputs (as defined in the Supported Models page). The issue has existed ever since we added support for image embedding inputs, i.e. #6613 (released in v0.5.5) ### Details Using image embeddings as an example: - For models that support image embedding inputs, the engine crashes when scattering the embeddings to `inputs_embeds` (mismatched shape) - For models that don't support image embedding inputs, the engine crashes when validating the inputs inside `get_input_embeddings` (validation fails). This happens because we only validate `ndim` of the tensor, but not the full shape, in input processor (via `MultiModalDataParser`). ### Impact - Denial of service by crashing the engine ### Mitigation - Use API key to limit access to trusted users. - Set `--limit-mm-per-prompt` to 0 for all non-text modalities to ban multimodal inputs, which includes multimodal embedding inputs. However, the model would then only accept text, defeating the purpose of using a multi-modal model. ### Resolution - https://github.com/vllm-project/vllm/pull/27204
AI coding agents often install or upgrade packages automatically in pip. A high vulnerability in a dependency can be pulled into a project through a normal install or update without a human reviewing the change, expanding the blast radius from a single package to every agent workspace that depends on it.
| Package | Affected range | Fixed version |
|---|---|---|
| vllmpip | >=0.5.5,<0.11.1 | 0.11.1 |
Fixed versions are reported by the source feed; confirm compatibility before updating.
Reported by GitHub Security Advisories (ghsa).
HOL Guard can help your team review package activity against supported protection paths.
Explore HOL GuardUpdate vllm to 0.11.1 if you use the affected versions. Test the change in a non-production environment first.
Local check
hol-guard supply-chain scanvLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs affects vllm (pip). Severity is high. ### Summary Users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct `ndim` but incorrect `shape` (e.g. hidden dimension is wrong), regardless of whether the model is intended to support such inputs (as defined in the Supported Models page). The issue has existed ever since we added support for image embedding inputs, i.e. #6613 (released in v0.5.5) ### Details Using image embeddings as an example: - For models that support image embedding inputs, the engine crashes when scattering the embeddings to `inputs_embeds` (mismatched shape) - For models that don't support image embedding inputs, the engine crashes when validating the inputs inside `get_input_embeddings` (validation fails). This happens because we only validate `ndim` of the tensor, but not the full shape, in input processor (via `MultiModalDataParser`). ### Impact - Denial of service by crashing the engine ### Mitigation - Use API key to limit access to trusted users. - Set `--limit-mm-per-prompt` to 0 for all non-text modalities to ban multimodal inputs, which includes multimodal embedding inputs. However, the model would then only accept text, defeating the purpose of using a multi-modal model. ### Resolution - https://github.com/vllm-project/vllm/pull/27204
AI coding agents often install or upgrade packages automatically in pip. A high vulnerability in a dependency can be pulled into a project through a normal install or update without a human reviewing the change, expanding the blast radius from a single package to every agent workspace that depends on it.
| Package | Affected range | Fixed version |
|---|---|---|
| vllmpip | >=0.5.5,<0.11.1 | 0.11.1 |
Fixed versions are reported by the source feed; confirm compatibility before updating.
Reported by GitHub Security Advisories (ghsa).
HOL Guard can help your team review package activity against supported protection paths.
Explore HOL Guard