Meta AI
Meta AI is an artificial intelligence assistant product developed by Meta, integrated into Meta's social platforms including Facebook, Instagram, WhatsApp, and Messenger. Users can access information retrieval, content creation, and image generation services through natural-language conversation. The product is built on Meta's proprietary large language models, such as the LLaMA series, and is designed to provide embedded AI interaction capabilities for social media contexts.
Core Facts
RAG（检索增强生成）架构由 Meta AI 研究团队在 2020 年提出
Both I-JEPA (the image version) and V-JEPA (the video version) were released and open-sourced by Meta AI
RAG (Retrieval-Augmented Generation) is an architectural paradigm proposed by Facebook AI Research in 2020.
Timeline (last 90 days)
The original Papers with Code website eventually stopped being maintained around 2024.
Meta AI acquired Papers with Code in 2021 and brought it in-house.
The official Meta AI account and Together AI will co-host an online deep-dive technical session at 12 noon Pacific Time on October 26.
Walmart launched Agent Gateway, allowing third-party AI assistants such as ChatGPT, Gemini, and Meta AI to connect directly to its product catalog to complete searches, price comparisons, and even add items to cart for users.
Ship SF speakers include Vercel CEO Guillermo Rauch, Shopify CEO Tobias Lütke, typesafeai CEO Diogo Almeida, Anthropic engineering lead Katelyn Lesse, and Meta AI VP of Engineering Prashant Ratanchandani.
FLORES devtest is the development test set in the FLORES-200 dataset released by Meta AI, with about 1,012 sentences per language, drawn from public texts such as Wikipedia and manually proofread.
The Meta AI blog once stated explicitly that open-source AI is the only viable path forward.
The official documentation does not explain what each of Muse Wonder's reasoning tiers specifically changes, nor does it give the differences between tiers in cost and latency.
Meta released the Muse Spark Open AI model, which it officially calls its most powerful US open-weight model to date.
Search Engine Land analyzed 9 million AI answers, covering nine major platforms ChatGPT, AI Overviews, Gemini, Perplexity, Copilot, Grok, Claude, and Meta AI, with the study covering more than 400 enterprise brands.
40 more timeline events
All Facts (20)
RAG（检索增强生成）架构由 Meta AI 研究团队在 2020 年提出
90%VerifiedBoth I-JEPA (the image version) and V-JEPA (the video version) were released and open-sourced by Meta AI
80%VerifiedRAG (Retrieval-Augmented Generation) is an architectural paradigm proposed by Facebook AI Research in 2020.
80%VerifiedThe concept of RAG was first formally proposed by Meta AI (formerly Facebook AI Research) in a 2020 paper.
80%VerifiedRAG (Retrieval-Augmented Generation) was proposed by Meta AI in 2020, compensating for the limitations of a model's parametric knowledge by dynamically retrieving external knowledge bases during LLM inference.
80%VerifiedRetrieval-Augmented Generation (RAG) was proposed by Meta AI in 2020.
80%VerifiedRAG was formally proposed by the Meta AI research team in 2020 in the paper "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks".
80%VerifiedThe core RAG architecture was first proposed by Patrick Lewis and others from Meta AI in a 2020 paper.
75%VerifiedMeta's AI smart glasses product line is represented by the Ray-Ban Meta series, launched in collaboration with Ray-Ban.
65%VerifiedPyTorch was released by Meta/Facebook's AI research team in 2016 and is known for its dynamic computation graph (eager execution) and Pythonic API.
65%VerifiedSAM (Segment Anything Model) is a foundation model released in 2023 capable of zero-shot segmentation of any object in an image; SAM 2, released in 2024, supports video segmentation.
65%VerifiedSAM (Segment Anything Model) is a model open-sourced by Meta that achieves zero-shot segmentation of arbitrary objects through large-scale pre-training on 11 million images and 1 billion masks.
65%VerifiedMeta's research division developed models such as SAM (Segment Anything Model) and ImageBind.
65%VerifiedThe RAG (Retrieval-Augmented Generation) architectural paradigm was proposed by Meta AI in 2020; its core idea is to retrieve relevant documents from external data sources as context before the model generates an answer.
65%VerifiedMeta has launched the AI-driven AI Studio tool and has deployed various generative AI features within Instagram and Facebook.
70%UnverifiedSegment Anything（SAM）是Meta推出的通用图像分割基础模型，能够零样本分割任意物体
90%UnverifiedX平台上的Grok AI主要面向付费订阅用户（Premium及以上），而Meta AI面向全体Threads用户
85%UnverifiedGAIA基准测试由Meta AI和HuggingFace等团队于2023年提出，包含400多个精心设计的问题
85%UnverifiedNaive RAG was proposed by Meta AI in 2020; its paradigm converts a question into a vector to retrieve relevant document fragments from a knowledge base, which are then fed into an LLM to generate an answer, in a single static pass.
70%UnverifiedThe additional AI usage credits in the Meta One package are likely designed around AI features such as Muse
60%