CVE-2025-29770: vLLM denial of service via outlines unbounded cache on disk
Impact The outlines library is one of the backends used by vLLM to support structured output (a.k.a. guided decoding). Outlines provides an optional cache for its compiled grammars on the local filesystem. This cache has been on by default in vLLM. Outlines is also available by default through the OpenAI compatible API server.
The affected code in vLLM is vllm/modelexecutor/guideddecoding/outlineslogitsprocessors.py, which unconditionally uses the cache from outlines. vLLM should have this off by default and allow administrators to opt-in due to the potential for abuse.
A malicious user can send a stream of very short decoding requests with unique schemas, resulting in an addition to the cache for each request. This can result in a Denial of Service if the filesystem runs out of space.
Note that even if vLLM was configured to use a different backend by default, it is still possible to choose outlines on a per-request basis using the guideddecodingbackend key of the extrabody field of the request.
This issue applies to the V0 engine only. The V1 engine is not affected.
Patches
https://github.com/vllm-project/vllm/pull/14837
The fix is to disable this cache by default since it does not provide an option to limit its size. If you want to use this cache anyway, you may set the VLLMV0USEOUTLINESCACHE environment variable to 1.
Workarounds
There is no way to workaround this issue in existing versions of vLLM other than preventing untrusted access to the OpenAI compatible API server.
References
Other sources
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. The outlines library is one of the backends used by vLLM to support structured output (a.k.a. guided decoding). Outlines provides an optional cache for its compiled grammars on the local filesystem. This cache has been on by default in vLLM. Outlines is also available by default through the OpenAI compatible API server. The affected code in vLLM is vllm/modelexecutor/guideddecoding/outlineslogitsprocessors.py, which unconditionally uses the cache from outlines. A malicious user can send a stream of very short decoding requests with unique schemas, resulting in an addition to the cache for each request. This can result in a Denial of Service if the filesystem runs out of space. Note that even if vLLM was configured to use a different backend by default, it is still possible to choose outlines on a per-request basis using the guideddecodingbackend key of the extrabody field of the request. This issue applies only to the V0 engine and is fixed in 0.8.0.
— MITRE
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
pip/vllmto a version that resolves this vulnerability.Fixed in 0.8.0 - Upgrade
Upgrade
vllmto a version that resolves this vulnerability.Fixed in 0.8.0 - Configuration
If you choose to use the outlines cache anyway, set the environment variable VLLM_V0_USE_OUTLINES_CACHE to 1; otherwise leave it disabled since unbounded on-disk caching can lead to denial of service.
vLLM outlines cache VLLM_V0_USE_OUTLINES_CACHE = 1 - Compensating control
Prevent untrusted access to the OpenAI compatible API server, since there is no other workaround in existing vLLM versions besides stopping untrusted clients from reaching the server.
Event History
Frequently Asked Questions
What is the severity of CVE-2025-29770?
CVE-2025-29770 is classified as a high-severity vulnerability due to its impact on the vLLM library.
How do I fix CVE-2025-29770?
To fix CVE-2025-29770, users should upgrade to vLLM version 0.8.0 or higher immediately.
What effect does CVE-2025-29770 have on the vLLM library?
CVE-2025-29770 can potentially lead to unintentional data exposure through the outlines caching mechanism.
Which versions of vLLM are affected by CVE-2025-29770?
CVE-2025-29770 affects all versions of vLLM from 0.0 up to, but not including, 0.8.0.
Is there a way to mitigate the risks associated with CVE-2025-29770 without upgrading?
While upgrading is the best solution, disabling the outlines caching feature can mitigate some risks associated with CVE-2025-29770.