Meta

Meta's multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default. safety classifier for LLM applications. Screens prompts and responses (text and images) against the MLCommons hazards taxonomy. Replaces Llama Guard 3 8B and Llama Guard 3 11B-Vision with a single model.

Meta's smallest prompt-injection detector — 22M parameters, sub-millisecond CPU inferenceRunning a model to get outputs — as opposed to training it. When you send a prompt to ChatGPT, that's inference. Inference is much cheaper than training per operation but adds up quickly at scale. Pricing pages almost always refer to inference costs (per million tokens, per request, etc.), not training costs., English-only. Targeted at high-throughput pipelines where the 86M variant's latency or cost is prohibitive.

Meta's prompt-injection detector — 86M parameter multilingual classifier built on mDeBERTa (not Llama architecture). Labels incoming prompts as benign or malicious to catch jailbreak attempts before they reach your primary LLM. Runs on CPU.

Base pretrained variant of Meta's Llama 4 Maverick. Frontier-class MoEA model architecture that splits the model into many smaller specialized "expert" networks, only activating a handful per input rather than running the whole model every time. The practical effect: you get the knowledge capacity of a big model with the compute cost of a much smaller one. Mistral Large 3 and Mistral Small 4 are both MoE models. with 17B active parametersIn a Mixture of Experts model, the number of parameters that actually run for any given input, as opposed to the total parameter count that's stored. Mistral Large 3, for example, has 675B total parameters but only 41B active per query — meaning it runs at roughly the cost of a 41B dense model while drawing on 675B worth of knowledge. out of 400B total, 128 experts, natively multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default., 1M context — designed for fine-tuning at scale rather than direct chat use.

Instruction-tuned variant of Meta's Llama 4 Maverick — Meta's largest open-weightA model where the trained weights are freely downloadable — you can run it yourself without contacting the creator. Llama, Mistral, Qwen, and Gemma are open-weight. Open-weight does not mean open-source: the training data and code often stay private. The license still governs what you can do with the weights, including whether you can use them commercially. release. 17B active / 400B total MoEA model architecture that splits the model into many smaller specialized "expert" networks, only activating a handful per input rather than running the whole model every time. The practical effect: you get the knowledge capacity of a big model with the compute cost of a much smaller one. Mistral Large 3 and Mistral Small 4 are both MoE models. with 128 experts, natively multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default., 1M context windowThe maximum amount of text the model can "see" at once — prompt plus prior conversation plus any documents you give it. Measured in tokens (which are roughly three-quarters of a word each). A 128K context window is about 96,000 words of input — roughly a 400-page book. Larger context windows let the model work with bigger documents but cost more to run.. Scores 80.5 on MMLUA broad knowledge test covering 57 subjects from law and medicine to mathematics and history. Scores are reported as percentage correct. A score around 85% is strong for a frontier model; above 90% is state-of-the-art. MMLU is probably the most-cited benchmark in AI model comparisons, though it has known weaknesses — models can memorize the questions, and the test reflects a specific cultural and academic context. Pro.

Base pretrained variant of Meta's Llama 4 Scout. Mixture-of-Experts architecture with 17B active parametersIn a Mixture of Experts model, the number of parameters that actually run for any given input, as opposed to the total parameter count that's stored. Mistral Large 3, for example, has 675B total parameters but only 41B active per query — meaning it runs at roughly the cost of a 41B dense model while drawing on 675B worth of knowledge. out of 109B total, natively multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default., designed for fine-tuning rather than direct chat use.

Meta's first mixture-of-experts and first natively multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default. open-weightA model where the trained weights are freely downloadable — you can run it yourself without contacting the creator. Llama, Mistral, Qwen, and Gemma are open-weight. Open-weight does not mean open-source: the training data and code often stay private. The license still governs what you can do with the weights, including whether you can use them commercially. model, with a 10-million-tokenThe basic unit of text a model reads and writes. Tokens are roughly three-quarters of a word in English — so 100 tokens is about 75 words. Models don't see letters or words directly; they see tokens. Pricing is almost always quoted per million tokens, and context windows are measured in tokens rather than words. context windowThe maximum amount of text the model can "see" at once — prompt plus prior conversation plus any documents you give it. Measured in tokens (which are roughly three-quarters of a word each). A 128K context window is about 96,000 words of input — roughly a 400-page book. Larger context windows let the model work with bigger documents but cost more to run. and an EU restriction business owners need to know about.

Meta's efficiency milestone — a 70-billion-parameter model that Meta claims matches their much larger 405B model on most tasks, at a fraction of the cost to run.

Meta's 11B Llama 3.2 Vision base modelA model straight out of pretraining, before any fine-tuning for chat or specific tasks. Base models predict the next token but don't follow instructions well — they'll continue your prompt rather than respond to it. Most people never use base models directly; they use the instruct-tuned or chat versions built on top. Useful mostly for researchers and people doing their own fine-tuning. — multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default. (text + image input) for research and fine-tuning. EU-domiciled entities are excluded from multimodal license rights. Most business use cases want the Instruct variant.

Meta's 11B Llama 3.2 Vision chat modelShorthand for an instruct-tuned model specifically designed for back-and-forth conversation rather than single-shot tasks. Chat models remember earlier turns in the conversation (within the context window) and respond in a conversational register. GPT-4, Claude, and most Llama Instruct variants are chat models. In practice, "chat model" and "instruct-tuned model" often mean the same thing. — multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default., handles text and image input for visual reasoning, document analysis, and chart reading. EU-domiciled entities are excluded from multimodal license rights.

Meta's 90B Llama 3.2 Vision base modelA model straight out of pretraining, before any fine-tuning for chat or specific tasks. Base models predict the next token but don't follow instructions well — they'll continue your prompt rather than respond to it. Most people never use base models directly; they use the instruct-tuned or chat versions built on top. Useful mostly for researchers and people doing their own fine-tuning. — multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default. flagship of the 3.2 generation, for research and fine-tuning. EU-domiciled entities are excluded from multimodal license rights. Most business use cases want the Instruct variant.

Meta's 90B Llama 3.2 Vision chat modelShorthand for an instruct-tuned model specifically designed for back-and-forth conversation rather than single-shot tasks. Chat models remember earlier turns in the conversation (within the context window) and respond in a conversational register. GPT-4, Claude, and most Llama Instruct variants are chat models. In practice, "chat model" and "instruct-tuned model" often mean the same thing. — multimodalA model that can handle more than one type of input or output — typically text plus images, sometimes plus audio or video. "GPT-4 Vision" and "Llama 3.2 11B Vision" are multimodal models that accept both text and images. A text-only model is called "unimodal" but nobody uses that term; text-only is the assumed default. flagship of the 3.2 generation, strong at complex visual reasoning and document analysis. EU-domiciled entities are excluded from multimodal license rights.

Base pretrained variant of Meta's smallest Llama 3.2 model. Text-only, 1.23B parameters, designed for on-deviceRunning a model directly on a consumer device — a laptop, a phone, a smart speaker — rather than in a data center. On-device inference keeps data private by never leaving the device, and works offline. Small models (under ~10B parameters, often quantized) can run on-device; larger models cannot yet. and edge deployment. For fine-tuning rather than direct chat use.

Meta's smallest instruction-tuned Llama 3.2 model. 1.23B parameters, text-only, designed for on-deviceRunning a model directly on a consumer device — a laptop, a phone, a smart speaker — rather than in a data center. On-device inference keeps data private by never leaving the device, and works offline. Small models (under ~10B parameters, often quantized) can run on-device; larger models cannot yet. chat, summarization, and prompt rewriting on phones or edge hardware. Commercial use allowed below the 700M MAU threshold.

Base pretrained variant of Meta's Llama 3.2 3B. Text-only, 3.21B parameters, designed for edge deployment and as a starting point for domain-specific fine-tuning.

Meta's Llama 3.2 3B Instruct — small instruction-tuned text model designed for mobile and edge AI assistants, with summarization, tool use, and multilingual dialogue capabilities. Commercial use allowed below the 700M MAU threshold.

Meta's Llama 3.1 405B base modelA model straight out of pretraining, before any fine-tuning for chat or specific tasks. Base models predict the next token but don't follow instructions well — they'll continue your prompt rather than respond to it. Most people never use base models directly; they use the instruct-tuned or chat versions built on top. Useful mostly for researchers and people doing their own fine-tuning. — the largest open-weightA model where the trained weights are freely downloadable — you can run it yourself without contacting the creator. Llama, Mistral, Qwen, and Gemma are open-weight. Open-weight does not mean open-source: the training data and code often stay private. The license still governs what you can do with the weights, including whether you can use them commercially. dense modelA model where every parameter is used for every input — the entire model runs on every token. Contrast with sparse or Mixture of Experts models, which activate only a fraction of the model per input. Dense models are simpler and more predictable; MoE models are more efficient at scale. ever released at the time. Base variant for research and large-scale fine-tuning; production deployments use the Instruct variant via hosted APIs.

Meta's Llama 3.1 405B chat modelShorthand for an instruct-tuned model specifically designed for back-and-forth conversation rather than single-shot tasks. Chat models remember earlier turns in the conversation (within the context window) and respond in a conversational register. GPT-4, Claude, and most Llama Instruct variants are chat models. In practice, "chat model" and "instruct-tuned model" often mean the same thing. — frontier-scale dense modelA model where every parameter is used for every input — the entire model runs on every token. Contrast with sparse or Mixture of Experts models, which activate only a fraction of the model per input. Dense models are simpler and more predictable; MoE models are more efficient at scale. with 128K context and strong multilingual reasoning. Practical access is through hosted APIAccessing a model by sending requests to the creator's (or a provider's) servers, typically pay-per-use. Hosted APIs handle all the operational work — scaling, hardware, uptime — in exchange for a per-token or per-request fee. Every closed-API model is hosted; many open-weight models are also available via hosted APIs from providers like Together, Fireworks, or Groq. providers; self-hosting requires a multi-GPUThe specialized chip that runs most AI models. Originally designed for 3D graphics, GPUs turned out to be excellent at the math AI requires. Nvidia dominates the AI GPU market; common datacenter models include the H100, H200, and B200. Running an AI model without a GPU is possible but painfully slow for anything but the smallest models. server cluster.

Llama 3.1 70B — the 70B Llama 3.1 base, widely used as a fine-tuning foundation (e.g. for Hermes).

Meta's Llama 3.1 8B base modelA model straight out of pretraining, before any fine-tuning for chat or specific tasks. Base models predict the next token but don't follow instructions well — they'll continue your prompt rather than respond to it. Most people never use base models directly; they use the instruct-tuned or chat versions built on top. Useful mostly for researchers and people doing their own fine-tuning. — 8B parameters, 128K context, multilingual. Base variant for fine-tuning; the Instruct variant has a full catalog entry.

Meta's small-but-serious open-weightA model where the trained weights are freely downloadable — you can run it yourself without contacting the creator. Llama, Mistral, Qwen, and Gemma are open-weight. Open-weight does not mean open-source: the training data and code often stay private. The license still governs what you can do with the weights, including whether you can use them commercially. model — fast, multilingual, and runs on a decent laptop with quantizationCompressing a model by reducing the numerical precision of its stored weights — for example, from 16-bit numbers to 4-bit numbers. The compressed model uses roughly a quarter of the memory and runs faster on most hardware, at the cost of slight accuracy loss. Quantization is what makes big models runnable on laptops — a 70B model in 4-bit quantization can fit on hardware that couldn't load the full-precision version., with a commercial license that works for almost any business.