Recommended Free Tools
Meta announced Llama 4 on April 5, 2025, introducing two downloadable open-weight models: Scout and Maverick. Both accept text and images, but they differ sharply in total size and context length. Behemoth was previewed as a model still in training—not released alongside them.
What Meta announced
Meta described Scout and Maverick as its first open-weight, natively multimodal models built with a mixture-of-experts (MoE) architecture. In an MoE model, only some of the model’s experts are active for a given input; the model card’s active-parameter count is therefore much smaller than its total parameter count. That distinction does not make the complete model footprint equivalent to a model with only that many parameters.
As an Amazon Associate I earn from qualifying purchases.
At launch, Meta offered Scout and Maverick through llama.com and Hugging Face. Meta also said its Llama 4-powered Meta AI could be tried in WhatsApp, Messenger, Instagram Direct, and on the web. Those are announcement-period availability statements, not confirmation that every option remains available in every region today.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →How Scout and Maverick differ
| Model | Active / total parameters | Context window | Meta’s stated deployment fit |
|---|---|---|---|
| Llama 4 Scout | 17B / 109B | 10 million tokens | Meta said Scout can fit on one NVIDIA H100 GPU with Int4 quantization. |
| Llama 4 Maverick | 17B / 400B | 1 million tokens | Meta said Maverick can fit on a single H100 host. |
The fit statements describe Meta’s specified deployment configurations, not a guarantee that either model will run easily on a typical desktop or laptop. Scout’s claim specifically depends on Int4 quantization. Maverick’s stated target is an H100 host, rather than a single H100 GPU.
#1 Best Overall
- CARDBOARD MONKENAUT — Get our best Gorilla Tag bundle yet with this Amazon exclusive deal. Purchase Meta Quest 3S to get exclusive items, including the Gorilla Space Program Suit and Helmet, plus 2,000 SHINY ROCKS.
- NO WIRES, MORE FUN — Break free from cords. Game, play and explore immersive worlds — untethered and without limits.
- 2X GRAPHICAL PROCESSING POWER — Enjoy lightning-fast load times and next-gen graphics for smooth gaming powered by the Snapdragon XR2 Gen 2 processor.
- EXPERIENCE VIRTUAL REALITY — Take gaming to a new level and blend virtual objects with your physical space to experience two worlds at once in your VR headset.
- 2+ HOURS OF BATTERY LIFE — Charge less, play longer and stay in the action with an improved battery that keeps up. *Based on the graphic performance of the Qualcomm Snapdragon XR2 Gen 2 platform vs the Meta Quest 2 platform.
Scout’s much longer context window may suit work involving large inputs, but context length alone does not establish which model will answer a particular question more accurately. The useful choice depends on the task, available hardware, latency needs, and how the model performs in your own evaluation.
What Llama 4 can do with text and images
The model card lists multilingual text and image input, with multilingual text and code output, for both Scout and Maverick. Meta identifies assistant-style chat and visual reasoning as intended uses for the instruction-tuned models; image recognition, image reasoning, captioning, and image question answering are also listed use cases.
Rank #2
- CARDBOARD MONKENAUT — Get our best Gorilla Tag bundle yet with this Amazon exclusive deal. Purchase Meta Quest 3 to get exclusive items, including the Gorilla Space Program Suit and Helmet, plus 2,000 SHINY ROCKS.
- NEARLY 30% LEAP IN RESOLUTION — Experience every thrill in breathtaking detail with sharp graphics and stunning 4K+ Infinite Display.
- NO WIRES, MORE FUN — Break free from cords. Game, play and explore in immersive worlds — untethered and without limits.
- 2X GRAPHICAL PROCESSING POWER — Enjoy lightning-fast load times and next-gen graphics for smooth gaming powered by the Snapdragon XR2 Gen 2 processor.
- EXPERIENCE VIRTUAL REALITY — Blend virtual objects with your physical space and experience two worlds at once in your VR headset.
The model card names 12 supported languages:
- Arabic
- English
- French
- German
- Hindi
- Indonesian
- Italian
- Portuguese
- Spanish
- Tagalog
- Thai
- Vietnamese
Although Meta says training included a broader language collection, the card places responsibility on developers to use the models safely when fine-tuning for languages beyond those listed as supported.
Free tools Windows power users keep installed
One-click scans. No signup required.
What the reported benchmarks do—and do not—show
Meta’s model card reports instruction-tuned Maverick at 80.5 on MMLU Pro and 69.8 on GPQA Diamond; Scout is listed at 74.3 and 57.2 on those same benchmarks. The card says these evaluations were run on BF16 models, even though quantized checkpoints are provided. These are developer-reported results on named tests, not an independent assessment of everyday product quality or proof that one model is better for every use.
Rank #3
- NO WIRES, MORE FUN — Break free from cords. Game, play, exercise and explore immersive worlds — untethered and without limits.
- 2X GRAPHICAL PROCESSING POWER — Enjoy lightning-fast load times and next-gen graphics for smooth gaming powered by the SnapdragonTM XR2 Gen 2 processor.
- EXPERIENCE VIRTUAL REALITY — Take gaming to a new level and blend virtual objects with your physical space to experience two worlds at once.
- 2+ HOURS OF BATTERY LIFE — Charge less, play longer and stay in the action with an improved battery that keeps up.
- 33% MORE MEMORY — Elevate your play with 8GB of RAM. Upgraded memory delivers a next-level experience fueled by sharper graphics and more responsive performance.
Meta’s launch announcement also made comparative performance claims about Scout and Maverick against other models. Those are Meta’s characterizations, not independent findings. Separately, NVIDIA reports TensorRT-LLM throughput above 40,000 tokens per second for Scout and 30,000 tokens per second for Maverick on B200 GPUs. These vendor-reported figures apply to the specified NVIDIA hardware and software stack; they should not be treated as expected speeds on other GPUs or ordinary computers.
Training data, cutoff, and disclosed footprint
The model card sets the pretraining data cutoff at August 2024. It describes a mixture of publicly available and licensed data, plus information from Meta products and services, including publicly shared Instagram and Facebook posts and people’s interactions with Meta AI. That disclosure does not establish that every user’s private content was included.
Rank #4
- NEARLY 30% LEAP IN RESOLUTION — Experience every thrill in breathtaking detail with sharp graphics and stunning 4K Infinite Display.
- NO WIRES, MORE FUN — Break free from cords. Play, explore and exercise in immersive worlds — untethered and without limits.
- 2X GRAPHICAL PROCESSING POWER — Enjoy lightning-fast load times and next-gen graphics for smooth gaming powered by the Snapdragon XR2 Gen 2 processor.
- EXPERIENCE VIRTUAL REALITY — Blend virtual objects with your physical space and experience two worlds at once.
- 2+ HOURS OF BATTERY LIFE — Charge less, play longer and stay in the action with an improved battery that keeps up.
Meta’s 2025 model card reports the following training figures:
- Approximately 40 trillion pretraining tokens for Scout and approximately 22 trillion for Maverick.
- 7.38 million cumulative H100-80GB GPU hours.
- An estimated 1,999 tons of location-based CO2-equivalent emissions.
- Zero tons of market-based emissions under Meta’s stated accounting approach.
The zero market-based figure reflects Meta’s clean-electricity matching and net-zero operational accounting statements; it does not mean the training used no electricity or had zero location-based emissions.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.License and deployment choices
“Open-weight” does not mean unrestricted or standard open-source licensing. The model card identifies the Llama 4 Community License Agreement as a custom commercial license and points users to Meta’s Acceptable Use Policy, including its out-of-scope uses. Anyone incorporating the weights into a product should review those terms for their specific use case.
Self-managed deployment is one route, subject to the model’s hardware requirements and license. Meta also announced Scout and Maverick for Amazon Bedrock and SageMaker JumpStart in 2025, which are hosted deployment paths; that announcement does not establish their current support status. Check the provider’s current model catalog and regional availability before planning a deployment.
What happened to Llama 4 Behemoth?
Behemoth was not one of the two downloadable models in the April 5, 2025 announcement. Meta described it as a larger teacher model that was still training. The available announcement information establishes its status at that time, but does not establish whether or when it later became available.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallQuick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
