Loading the catalogue…
Desktop apps · Element Labs, Inc. · US
LM Studio
US CLOUD Act exposureProprietary
STAV REVIEW · LAST REVIEWED 06/21/2026
“Architecturally air-gapped inference with a strong vendor privacy commitment, undercut by two unresolved GGUF supply-chain disclosures and new connectivity features the privacy policy has not yet caught up with.”
Good for
- Teams requiring fully local, air-gappable inference where no prompt or completion data can leave the device by design
- Developers integrating a headless local inference backend via the OpenAI-compatible REST API or lmstudio-js SDK without changing existing application code
- Enterprises deploying open-weight models on their own hardware who need on-premises control and want a weekly-release-cadence desktop runtime
- Security-conscious teams using Stav's audit-logging layer on top, since LM Studio does not centralise session or model-selection metadata itself
Watch out for
- Two unresolved supply-chain disclosures: GGUF chat-template poisoning (July 2025, vendor declined responsibility) and six llama.cpp GGUF parser vulnerabilities including one rated critical (May 2026, no CVEs assigned, no confirmed remediation) — restrict model downloads to a vetted internal registry until these are formally closed
- LM Link remote-access and the Locally iPhone/iPad app, both launched June 2026, introduce opt-in connectivity flows whose relay topology and data-flow profile are not covered by the current privacy policy (effective March 2025) — block by policy until Element Labs publishes an updated policy
- No in-app opt-out for update-check telemetry (OS version, app version, IP address) and no opt-out for anonymised model-search queries; both flows are always-on with no documented suppression mechanism
Editorial assessment, not legal advice. Stav's risk ratings, scores, and verdicts are our own analysis of publicly available information and may be incomplete or out of date. Verify independently before making compliance or procurement decisions.
MODEL REQUIREMENTS
What LM Studio needs from a model.
Required
- Open weights
- Streaming
Recommended
- Long context (≥128k)
- Structured output
Data flow & telemetry
How LM Studio moves data between the client, the inference endpoint and any vendor-owned services.
Data flow
- Direction
- local_only
- Stores prompts
- No
Telemetry
- Collects telemetry
- true
- Privacy policy
- https://lmstudio.ai/app-privacy
Vendor & licensing
- Developer
- Element Labs, Inc.
- Vendor country
- US
- Vendor type
- startup
- Funding status
- unknown
- Runtime type
- desktop_native
- Open source
- No
- Licence
- Proprietary
- Licence risk
- commercial_only
- CLOUD Act exposure
- Yes
Security & maintenance health
Security profile
- Notes
- No CVEs on record for LM Studio directly across multiple research runs; NVD API was unavailable for fresh verification (HTTP errors on all keyword attempts in this run) — manual NVD/GHSA check is advisable before high-security approvals. Two unresolved supply-chain disclosures are material: (1) Pillar Security 'Poisoned GGUF Templates' (July 2025, medium severity) — vendor declined responsibility, no formal mitigation published; (2) six llama.cpp GGUF parser vulnerabilities (oss-security 2026-05-15, V-01 through V-06; V-01 rated critical by the author, integer overflow enabling out-of-bounds memory reads) — no CVEs assigned; vendor's v0.4.13, v0.4.15, and v0.4.16 'Security hardening' changelog entries may partially address these but are unconfirmed. Binary code-signing status (Apple notarisation, Windows Authenticode) could not be verified from public metadata.
- Bug Bounty
- false
- Auto Update
- true
- Cve Count 90d
- 0