110 Total
62 High severity
46 Medium severity
2 Low severity

Key Facts

What is Meta's total aggregate liability limited to?
Meta limits its total aggregate liability to the greater of the amounts paid by the user to Meta for use of the Services in the preceding twelve-month period or one thousand dollars.
Is Meta's liability limited to the greater of amounts paid in the preceding twelve-month period or one thousand dollars?
Meta limits its total aggregate liability to the greater of the amounts paid by the user to Meta for use of the Services in the preceding twelve-month period or one thousand dollars.
Will neither party nor its affiliates be liable for loss of use or loss of profits?
Neither party nor its affiliates, licensors, officers, and employees will be liable for loss of use, loss of profits, lost or inaccurate data or content, interruption or loss of business, or any indirect or consequential damages of any kind.
Will Meta indemnify users of Paid Services against damages arising from third-party claims?
Meta will indemnify users of Paid Services against damages finally awarded by a court of competent jurisdiction arising from third-party claims that the user's use of the Paid Services, Meta's use of training data, or unmodified Outputs infringe any intellectual property rights.
What may Meta do at its sole discretion?
Meta may, at its sole discretion, add, change, or remove functionality, features, or other aspects of the Services, and throttle, limit, suspend, or terminate user access at any time without notice.
Can Meta add, change, or remove functionality, features, or other aspects of the Services?
Meta may, at its sole discretion, add, change, or remove functionality, features, or other aspects of the Services, and throttle, limit, suspend, or terminate user access at any time without notice.
Can Meta throttle, limit, suspend, or terminate user access without notice?
Meta may, at its sole discretion, add, change, or remove functionality, features, or other aspects of the Services, and throttle, limit, suspend, or terminate user access at any time without notice.
Can users reverse engineer, decompile, disassemble, or otherwise attempt or assist others to derive the underlying models, algorithms, weights, or source code of the Services?
Meta prohibits users from reverse engineering, decompiling, disassembling, or otherwise attempting or assisting others to derive the underlying models, algorithms, weights, or source code of the Services.
What must users not submit to the Unpaid Services?
Meta requires that users must not submit sensitive, confidential, or personal information to the Unpaid Services.
Can users submit sensitive, confidential, or personal information to the Unpaid Services?
Meta requires that users must not submit sensitive, confidential, or personal information to the Unpaid Services.
Stay ahead of the changes
Track Meta and get the diff the day its terms change.
Summary

This document sets the rules for using Meta's Llama API. If you use the free tier, Meta can use everything you input and every output you receive to improve its AI systems, so you should not submit private or sensitive information. Meta can cut off your access at any time without warning, its financial liability to you is capped at $1,000 or what you have paid in the past year, and most disputes must go through arbitration rather than court.

Analysis

The Llama API Terms of Service establish the rights and obligations governing developer access to Meta's Llama API services. Meta retains sole discretion to modify, throttle, suspend, or terminate access to the Services at any time without notice, while users are prohibited from reverse engineering the Services, sharing API keys without written permission, and using Outputs to train competing AI models without prior written authorization. Users of the Unpaid Services grant Meta a right to use their Inputs and Outputs to train and improve Meta's AI models, products, and services, and must not submit sensitive or personal information through those services; users of Paid Services receive indemnification from Meta for qualifying third-party intellectual property claims. Meta disclaims all warranties, excludes all indirect and consequential damages, and caps its aggregate liability to the greater of amounts paid in the preceding twelve months or $1,000, with US users required to arbitrate most disputes except those involving intellectual property rights.

What this means for you

Developers using the Unpaid Services should be aware that their Inputs and Outputs are used by Meta to train and improve its AI models, and they are contractually required not to submit sensitive, confidential, or personal information through those services. Users of Paid Services benefit from Meta's indemnification against third-party intellectual property claims for qualifying uses, but this protection does not extend to Unpaid Services users. Meta can suspend or terminate access at any time without notice at its sole discretion, and any financial recovery from Meta is capped at the greater of amounts paid in the past twelve months or $1,000. Developers may preserve their right to litigate intellectual property claims in court, as those claims are expressly excluded from the mandatory arbitration requirement.

Institutional Analysis
Stay ahead of the changes

Institutional analysis available with Insight

Which mapped governance frameworks each document engages, tied to the specific provisions that engage them.

7 important changes detected

8 versions captured · Last updated: August 2026

What changed Meta expanded the Meta Model API Terms of Service to introduce tiered service models (Discounted and Standard Models), added expanded definitions of Inputs to include libraries, code, and databases accessed through connectors, clarified intellectual property ownership between parties, and added explicit AI disclosure and labeling obligations for developers. The terms now specify that developers must comply with applicable AI classification and provenance-marking requirements when deploying outputs to end users, and introduced new license restrictions governing downloadable materials and third-party software included in the services.
Why this matters The updated terms establish new obligations for developers integrating Meta's AI models into their products. The agreement now requires developers to comply with applicable AI labeling, disclosure, provenance-marking, and system classification requirements under applicable law, with responsibility falling on the developer to implement these disclosures to end users. The expanded definition of Inputs now explicitly includes code, libraries, databases, and any content accessed through connectors or integrations, which may broaden what user-submitted content Meta claims the right to process for safety and security purposes.
View full change record →
What changed Meta added a 'Contact us' link to the header of their Llama API Terms of Service, detected on July 22, 2026. The substantive governance language of the terms—covering access to services during a limited preview period—remains unchanged. This is a formatting or navigation modification to the document structure, not a change to the rights, obligations, or operational rules governing API use.
Why this matters This change adds a 'Contact us' link to the header of the Llama API Terms of Service. The substantive terms governing access and use of the services remain unchanged. This is a navigation or structural modification with no material impact on API usage rights, restrictions, or obligations.
View full change record →

July 9, 2026 medium

Meta rebranded its Llama API service to 'Meta Model API' and restructured the terms significantly. The updated agreement expands the scope from Llama-specific models to broader 'Meta Models', removes the …

View change record →
July 7, 2026 low

Meta updated the release date format in the Llama API Terms of Service header on July 7, 2026. The previous version stated 'Release Date: April 29, 2025' with a comma …

View change record →
May 29, 2026 low

Meta updated its Llama API Terms of Service on May 29, 2026, making a minor stylistic change to a sentence describing when Meta may retain user content. The updated version …

View change record →
May 21, 2026 medium

Meta updated its Llama API Terms of Service on May 21, 2026 to add a third reason it may retain user content. Previously, Meta stated it could retain content for …

View change record →
May 15, 2026 low

Meta removed the 'Llama API Service status: Available' line and 'Contact us' text from the header of its Llama API Terms of Service on May 15, 2026. The Terms themselves …

View change record →
Featured, High severity

Complete Provision Index

Every distinct legal provision identified in this document. Featured provisions appear above with analysis.

110 provisions
12 featured
24 clause types
62 high severity
Content Moderation 1 1 high
Disclosure and Transparency Requirements 1 1 high
Refunds & Chargebacks 1 1 high
Targeting and Audience Restrictions 1 1 high
Data Sharing 1
Stay ahead of the changes

Monitoring

Meta has updated this document before. Monitor includes same-day alerts, structured change summaries, and monitoring for up to 20 platforms.

Stay ahead of the changes

Governance Intelligence

Need provision-level monitoring and regulatory mapping? Insight includes governance timelines, drift analysis, and full provision tracking.

Cross-platform context

See how other platforms handle Account termination auto-terminates these Terms and similar clauses.

Compare across platforms →

Mapped Governance Frameworks

CFAA
United States Federal
View official text ↗
DMCA
United States Federal
View official text ↗
DSA
European Union
View official text ↗
ePrivacy Directive
European Union
View official text ↗
FTC Act Section 5
United States Federal
View official text ↗

Related Analysis

Privacy · May 8, 2026
Meta Removed 438 Sentences From Its Privacy Policy. Here Is What Disappeared.

ConductAtlas detected a major restructuring of Meta’s privacy policy that removed detailed consumer rights disclosures and relocated them t…

Archival ProvenanceSource & Archival Record
Last Captured August 7, 2026 00:10 UTC
Capture Method Automated scheduled archival capture
Document ID CA-D-000778
Version ID CA-V-005548
SHA-256 21f2a2feaa022154750b677ee97104d8f8717bc3fa7131db04b776d36689c1b4
✓ Snapshot stored ✓ Text extracted ✓ Change verified ✓ Hash verified

Governance Monitoring

Monitor governance changes across the platforms you rely on.

Structured alerts for policy changes, governance events, and provision updates across 352+ platforms.

Start monitoring → Compare plans