Llama 4
Product status: Status not recorded
No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.
No status check date recorded
Meta's Scout and Maverick multimodal weight releases for text/image input and text/code output, under the Llama 4 Community License. No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.
Decision notes
Official-document research
Based on the cited documentation. We did not use the product, purchase a plan, or run quality, speed or reliability tests in this review.
Consider it for
- Teams qualifying for the model license
- Operators evaluating a specific multimodal checkpoint
Choose another approach if
- EU developers assuming an unrestricted grant
- Guaranteed 10M context on every variant or host
Pricing and conditions
Documented terms
The community license offers a limited royalty-free grant subject to AUP, redistribution and separate commercial-license conditions. This is not a free API subscription. Compute, model storage, inference infrastructure, operations and any provider fees remain separate. No hosted per-token price, currency, tax or regional quote was confirmed; use the exact checkpoint/precision configuration rather than a universal hardware-cost claim.
Pricing source read:
Pricing or terms sourceCode, model and service licensing
Llama 4's custom license is not blanket OSI licensing. Redistribution requires agreement/NOTICE and Built with Llama branding; distributed models improved with its materials or outputs require a Llama-prefixed name. The release-date prior-month 700M MAU threshold can require a separate Meta license, with a narrow integrated-end-user exception. Its AUP withholds the multimodal §1(a) grant from EU-domiciled individuals/EU-principal-business companies; end users of an incorporating product/service are excepted. Use restrictions, input consents and output rights remain separate. A self-hosted checkpoint does not determine external tools' or hosts' training/retention terms.
Basis and limits of the recorded score
The existing 4.5 score is retained; its original scoring evidence is unavailable. This documentation review did not measure performance or compare products and did not re-score it.
Overview
Llama 4 covers the April 2025 Scout and Maverick releases. The public card specifies text/image input and text/code output, with Scout's 10M-token and Maverick's 1M-token context limits. Those are model specifications, not evidence that every serving stack or workload supports the maximum, or that long-input answers are complete. Supported languages and visual-task limits also need checking.
Select checkpoint, precision and hardware together. The repository distinguishes full precision from FP8/Int4 examples, so one-GPU statements cannot be applied to all formats or models. Public cards and the complete license are readable even though access to files requires an application. This review neither downloaded the files nor inferred retirement from Meta's changing homepage.
Sources
The factual claims in this review were read from the pages below. Each date is when an editor last opened the source, not when the product changed.
- Vendor documentation Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
- Vendor documentation Checked 2026-10-15
Pros / Cons
Pros
- Published multimodal cards
- Weights can run in an authorized deployment
- Separate Scout and Maverick specifications
Cons
- EU multimodal grant restriction
- Substantial format-dependent resources
- Custom-license and AUP obligations
Who it's for
Evaluating authorized image-and-text applications
Testing a pinned checkpoint on suitable infrastructure
Bottom line
Choose Llama 4 only after confirming the deployment jurisdiction, checkpoint license and resource budget.
Key Features
Tags
Frequently asked questions
Is Llama 4 currently available?
No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.
What is Llama 4?
Llama 4 is a Conversational AI tool developed by Meta. Meta's Scout and Maverick multimodal weight releases for text/image input and text/code output, under the Llama 4 Community License.
What costs and terms should I check for Llama 4?
The community license offers a limited royalty-free grant subject to AUP, redistribution and separate commercial-license conditions. This is not a free API subscription. Compute, model storage, inference infrastructure, operations and any provider fees remain separate. No hosted per-token price, currency, tax or regional quote was confirmed; use the exact checkpoint/precision configuration rather than a universal hardware-cost claim.
Is Llama 4 open source?
Llama 4's custom license is not blanket OSI licensing. Redistribution requires agreement/NOTICE and Built with Llama branding; distributed models improved with its materials or outputs require a Llama-prefixed name. The release-date prior-month 700M MAU threshold can require a separate Meta license, with a narrow integrated-end-user exception. Its AUP withholds the multimodal §1(a) grant from EU-domiciled individuals/EU-principal-business companies; end users of an incorporating product/service are excepted. Use restrictions, input consents and output rights remain separate. A self-hosted checkpoint does not determine external tools' or hosts' training/retention terms.
What are the best alternatives to Llama 4?
Notable Conversational AI alternatives include ChatGPT, Claude, Gemini. Browse all 67 tools in our Conversational AI category for a full overview.
Was this listing useful?
Basic Info
- Pricing
- Free
The community license offers a limited royalty-free grant subject to AUP, redistribution and separate commercial-license conditions. This is not a free API subscription. Compute, model storage, inference infrastructure, operations and any provider fees remain separate. No hosted per-token price, currency, tax or regional quote was confirmed; use the exact checkpoint/precision configuration rather than a universal hardware-cost claim.
- Editorial score
- ★ 4.5 / 5 How we score
- Platforms
- WeightsSelf-hosted
- Last verified
- Not recorded
