Text Generation ModelDeprecated

Llama 3 70B

Llama 3 70B is a text generation model from Meta, instruction-tuned for dialogue with an 8,192-token context window.

Start Building with Llama 3 70B View All Models

Publisher

Open-weight dialogue model from Meta

Llama 3 70B is a 70-billion-parameter large language model developed and released by Meta as part of the Llama 3 model family, which also includes an 8B variant. The model is available in both pretrained and instruction-tuned forms, with the instruction-tuned version optimized specifically for dialogue and conversational use cases. It is an open-weight model, meaning the weights are publicly released for research and commercial use under Meta's Llama 3 license.

The instruction-tuned version of Llama 3 70B was developed with an emphasis on helpfulness and safety, making it suited for chat applications, question answering, and general-purpose text generation tasks. On MindStudio, this model is served via Groq, which provides hardware-accelerated inference. Note that this model has been marked as deprecated in the MindStudio catalog, so developers building new applications may want to consider a current alternative.

Capabilities

What Llama 3 70B supports

Instruction Following

The instruction-tuned variant is optimized for dialogue and follows multi-turn conversational prompts, making it suitable for chat and assistant-style applications.

Text Generation

Generates coherent long-form text across a range of formats including summaries, explanations, and creative writing, up to 8,192 tokens per response.

Question Answering

Responds to factual and open-ended questions using knowledge encoded during pretraining on a large multilingual corpus.

Code Assistance

Capable of generating, explaining, and debugging code across common programming languages as a general capability of the 70B model class.

Reasoning Tasks

Handles multi-step reasoning problems including logical inference and structured problem-solving within its 8,192-token context window.

Ready to build with Llama 3 70B?

Get Started Free

FAQ

Common questions about Llama 3 70B

What is the context window for Llama 3 70B?

The context window is 8,192 tokens, which also serves as the maximum response size for this model.

Who developed Llama 3 70B?

Llama 3 70B was developed and released by Meta. On MindStudio, it is served through Groq's inference infrastructure.

Is this model still available to use?

Llama 3 70B (llama3-70b-8192) is currently marked as deprecated in the MindStudio catalog. Developers starting new projects may want to select an active model instead.

What is the knowledge cutoff for Llama 3 70B?

Meta has not published an exact knowledge cutoff date in the metadata available here. Based on public information, Llama 3 models were trained on data with a cutoff of approximately early 2023.

Does Llama 3 70B support image or video inputs?

No. This version of Llama 3 70B is a text-only model and does not support image or video analysis.

Resources