Skip to content

litellm

47 advisories in the AI Sec Watch database name litellm (PyPI). litellm (PyPI) has declared an LLM component as a dependency since 2023-07-27. The most recent advisory that names litellm and states a fix gives 1.95.0.

Library to easily interface with LLM API providers

The first release to declare it was version 0.1.0.

PyPILatest 1.109.1First release 2023-07-27Registry pageRepositoryChecked 2026-11-14

Advisories

Advisories that name litellm as affected and state PyPI as its ecosystem. For a version-exact check of your own dependencies, use Stack Check.

Time to fix: a median of −24 days from disclosure to the release of the fixed version, over 39 advisories with a dated fix (fastest −163, slowest +9 days). A negative count means the fix was released before the advisory's disclosure date.

AdvisorySeverityAffectedFixed inFix releasedPublished
GHSA-g5ff-637f-6q2m: LiteLLM: internal_user_viewer arbitrary local file read via `vertex_ai_credentials`High< 1.95.01.95.02026-08-02 (−86 d)2026-10-27
GHSA-hhww-mrg2-969h: LiteLLM vulnerable to reflected XSS in /sso/debug/callbackMedium>= 1.65.5, < 1.85.01.85.02026-05-17 (−163 d)2026-10-27
CVE-2026-114612LiteLLM admin access via forged credential replay enables MCP stdio commandsCritical>= 1.91.0, < 1.100.41.100.42026-09-29 (−23 d)2026-10-22
CVE-2026-59823LiteLLM Proxy server-side request forgery through user_config api_baseHigh<= 1.83.81.83.92026-04-17 (−152 d)2026-09-16
CVE-2026-84377LiteLLM proxy credential exposure through unchecked api_base redirectionMedium< 1.88.61.88.62026-08-09 (−24 d)2026-09-02
CVE-2026-37004LiteLLM unauthenticated command execution via template injectionCritical< 1.83.71.83.72026-04-13 (−136 d)2026-08-27
CVE-2026-59822LiteLLM authentication bypass through MCP Streamable HTTP endpointHigh< 1.84.01.84.02026-05-14 (−55 d)2026-07-08
CVE-2026-59821LiteLLM code injection through Custom Code Guardrails create and updateLow< 1.82.01.82.02026-03-01 (−129 d)2026-07-08
CVE-2026-59820LiteLLM path traversal in Skills archive extraction via uploaded ZIPHigh< 1.83.71.83.72026-04-13 (−86 d)2026-07-08
CVE-2026-59819LiteLLM file read through /health/test_connection endpointLow< 1.83.101.83.102026-04-19 (−80 d)2026-07-08
CVE-2026-12796GHSA-w2mh-qq9q-453x: BerriAI litellm: SSO Login Does Not Invalidate Previous UI Session TokensLow<= 1.82.22026-06-21
CVE-2026-12797GHSA-p897-vf7j-f5h8: BerriAI litellm has Security Feature Bypass in BannedKeywords and AzureContentSafety Guardrails via call_type Mismatch on Async EndpointsLow<= 1.82.52026-06-21
CVE-2026-12798GHSA-c693-x898-5g4h: BerriAI litellm has SSRF via Unvalidated spec_path URL in MCP OpenAPI Spec LoaderLow<= 1.82.22026-06-21
CVE-2026-12799GHSA-m2v5-74w2-qhcj: BerriAI litellm: UI User Enumeration leads to System-Wide Information DisclosureLow<= 1.82.22026-06-21
CVE-2026-12795GHSA-j37q-q7p9-vpwm: LiteLLM: SSO Debug Flow Has Improper AuthenticationMedium<= 1.82.22026-06-21
CVE-2026-12773GHSA-4jcj-7x88-m979: LiteLLM: MCP Proxy Has Improper AuthenticationMedium< 1.84.01.84.02026-05-14 (−38 d)2026-06-21
CVE-2026-12770GHSA-6qr3-3g89-m4jj: LiteLLM: Admin Key Handler Has Improper AuthorizationLow<= 1.63.12026-06-21
CVE-2026-12772GHSA-mf52-j94g-746m: LiteLLM: PROXY_ADMIN database API Key Generator Has Insufficient Session ExpirationLow<= 1.82.22026-06-21
CVE-2026-12771GHSA-qmf3-4767-5fg3: LiteLLM: M2M JWT Handler Has Improper AuthorizationLow<= 1.82.22026-06-21
CVE-2026-49468GHSA-4xpc-pv4p-pm3w: LiteLLM: Authentication Bypass via Host Header InjectionCritical< 1.84.01.84.02026-05-14 (−33 d)2026-06-16
CVE-2026-47102LiteLLM self-service user update allows changing own user_role via /user/updateHigh< 1.83.101.83.102026-04-19 (−32 d)2026-05-21
CVE-2026-47101LiteLLM API key creation bypasses role-based route restrictionsHigh< 1.83.141.83.142026-04-26 (−25 d)2026-05-21
CVE-2026-42271LiteLLM command execution through MCP test endpointsCritical>= 1.74.2, < 1.83.71.83.72026-04-13 (−25 d)2026-05-08
CVE-2026-42208CVE-2026-42208: LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.81.16 to before…Critical>= 1.81.16, < 1.83.71.83.72026-04-13 (−25 d)2026-05-08
CVE-2026-42203LiteLLM code execution through prompt template rendering in /prompts/testCritical>= 1.80.5, < 1.83.71.83.72026-04-13 (−25 d)2026-05-08
GHSA-v4p8-mg3p-g94g: LiteLLM: Authenticated command execution via MCP stdio test endpointsHigh>= 1.74.2, < 1.83.71.83.72026-04-13 (−12 d)2026-04-25
GHSA-r75f-5x8p-qvmc: LiteLLM has SQL Injection in Proxy API key verificationCritical>= 1.81.16, < 1.83.71.83.72026-04-13 (−11 d)2026-04-24
GHSA-xqmj-j6mv-4862: LiteLLM: Server-Side Template Injection in /prompts/test endpointHigh>= 1.80.5, < 1.83.71.83.72026-04-13 (−11 d)2026-04-24
CVE-2026-40217LiteLLM remote code execution at /guardrails/test_custom_codeHigh>= 1.81.8, < 1.83.101.83.102026-04-19 (+9 d)2026-04-10
GHSA-69x8-hrgq-fjj8: LiteLLM: Password hash exposure and pass-the-hash authentication bypassHigh< 1.83.01.83.02026-03-31 (−8 d)2026-04-08
CVE-2026-35030GHSA-jjhc-v7c2-5hh6: LiteLLM: Authentication bypass via OIDC userinfo cache key collisionCritical< 1.83.01.83.02026-03-31 (−3 d)2026-04-03
CVE-2026-35029GHSA-53mr-6c8q-9789: LiteLLM: Privilege escalation via unrestricted proxy configuration endpointHigh< 1.83.01.83.02026-03-31 (−3 d)2026-04-03
GHSA-5mg7-485q-xm76: Two LiteLLM versions published containing credential harvesting malwareCritical>= 1.82.7, <= 1.82.82026-03-25
CVE-2024-10188LiteLLM denial of service through unsafe parsing of user inputMedium< 1.53.1.dev11.53.1.dev12024-12-03 (−107 d)2025-03-20
CVE-2025-0330GHSA-879v-fggm-vxw2: LiteLLM Has a Leakage of Langfuse API KeysHigh<= 1.52.12025-03-20
CVE-2025-0628GHSA-fjcf-3j3r-78rp: LiteLLM Has an Improper Authorization VulnerabilityHigh< 1.61.151.61.152025-02-23 (−25 d)2025-03-20
CVE-2024-9606GHSA-g5pg-73fc-hjwq: LiteLLM Reveals Portion of API Key via a Logging FileHigh< 1.44.121.44.122025-03-20
CVE-2024-8984GHSA-fh2c-86xm-pm2x: LiteLLM Vulnerable to Denial of Service (DoS) via Crafted HTTP RequestHigh< 1.56.21.56.22024-12-27 (−83 d)2025-03-20
CVE-2024-6825GHSA-53gh-p8jc-7rg8: LiteLLM Vulnerable to Remote Code Execution (RCE)High>= 1.40.3.dev2, <= 1.40.122025-03-20
CVE-2024-6587litellm server-side request forgery through api_base in chat completionsHigh< 1.44.81.44.82024-09-13
CVE-2024-5710GHSA-qqcv-vg9f-5rr3: litellm vulnerable to improper access control in team managementMedium< 1.40.151.40.152024-06-27
CVE-2024-5751GHSA-gppg-gqw8-wh9g: litellm vulnerable to remote code execution based on using eval unsafelyCritical< 1.40.161.40.162024-06-27
CVE-2024-4888BerriAI litellm arbitrary file deletion through /audio/transcriptions endpointHigh< 1.35.361.35.362024-06-06
CVE-2024-4890GHSA-8j42-pcfm-3467: SQL injection in litellmMedium<= 1.27.142024-06-06
CVE-2024-5225GHSA-h6m6-jj8v-94jj: SQL injection in litellmMedium< 1.40.01.40.02024-06-06
CVE-2024-4264GHSA-7ggm-4rjg-594w: litellm passes untrusted data to `eval` function without sanitizationHigh<= 1.28.112024-05-18
CVE-2024-2952GHSA-46cm-pfwv-cgf8: LiteLLM has Server-Side Template Injection vulnerability in /completions endpointCritical< 1.34.421.34.422024-04-10

Dependencies of the latest release

As declared in PyPI metadata for version 1.109.1. Optional extras are listed with their extra name.

Tracked packages that depend on it

Among the packages in the registry; not every dependent on PyPI.